This study evaluated brain activity over a 15-min CPT-IP vigil in manic/mixed patients with bipolar disorder and healthy individuals. As predicted, the BP group evidenced a decrement in CPT-IP sensitivity over time that was steeper than in the HC group. Consistent with the previous literature2,6
, the BP group suffered from an increasingly pronounced sustained attention deficit despite relatively good baseline performance. However, the effect was only partially confirmed by a marginally significant Group x Time interaction, so despite consistency with the previous literature, this result should be considered preliminary until it can be further replicated. This deficit was not the result of group differences in response bias, which is sensitive to fatigue and motivation, or response speed2
The CPT-IP initially activated prefrontal cortex as demonstrated in voxel-wise analysis and confirmed in ROI analysis relative to baseline activation. As expected for a CPT, regions of DLPFC (SFG and IFG) were strongly activated in both study groups at vigilance period 1. Beyond vigilance period 1, the HC group continued to activate IFG at vigilance periods 2 and 3, while the BP group failed to activate frontal cortex relative to baseline levels. These results suggest that while both groups initially activated frontal control mechanisms early in the vigil this activation rapidly diminished in the BP group after five minutes.
Consistent with the second prediction, the BP group not only overactivated bilateral amygdala relative to the HC group as in previous studies22
, but the HC group deactivated amygdala, suggesting that patients with bipolar disorder have functional abnormalities in both the magnitude and direction of activation. The amygdala was the only structure demonstrating bilateral Group effects (BP>HC) and the only structure in which activation consistently differed relative to baseline (BP>baseline>HC). This finding may represent abnormally increased emotional input to the anterior-limbic network in bipolar disorder. Although the Group effect was also qualified by an interaction with Time in right amygdala, even here the group effect was obvious and significant.
Unlike the first and second prediction, the third prediction was only partially confirmed. Although both study groups demonstrated less left SFG activation over time, the BP group showed no less activation than the HC group as evidenced by the lack of an interaction. Additionally, no significant interactions were identified for IFG. These findings suggest that left SFG activation, in particular, decreases over extended vigils, implicating dorsolateral prefrontal cortex in the CPT performance of both healthy and psychiatric samples17
Although no distinct differences in prefrontal cortical activation were identified between the groups as expected on the rationale that the BP group would have less cortical control of attention as a consequence of increased emotional input to the system by amygdala overactivation, other regions within the VLPFC-subcortical circuit showed differential group activation over time. In particular, both left striatum and bilateral thalamic activation differed by group over time as indicated by significant or marginally significant interactions. Within the striatum, the BP and HC groups progressively decreased and increased activation over time, respectively. Within the thalamus, the BP group decreased activation over time relative to a flat temporal pattern in the HC group. Blumberg et al.19
also reported striatal (i.e., putamen) and thalamic dysfunction in adolescents in a primarily depressed state of bipolar disorder on a test of selective attention (i.e., a Stroop task), but activation was increased relative to healthy controls, possibly indicating a differential effect of mood state (depression vs. mania) or attentional task (Stroop vs. CPT-IP), or a developmental change from adolescence to adulthood19
. Although the significance of these findings is uncertain, they may suggest that subcortical structures within the VLPFC-subcortical circuit, and anterior-limbic network more generally, are the primary regions of abnormal attentional activation across adolescence and early adulthood during acute states of bipolar disorder. It is also possible that there is a specific loss of prefrontal control of subcortical iterations in bipolar disorder as previously demonstrated by attenuated activation in ventral prefrontal cortices relative to healthy controls18
that the CPT-IP is less sensitive to than are some other tasks.
In the BP group subcortical activation decrements occurred as attention waned, while in the HC group activation was more stable or even increased over time, possibly representing a compensatory mechanism to maintain attention. Perhaps, in the absence of increased VLPFC activation to monitor and regulate emotional output by the amygdala as demonstrated in euthymia21
, subcortical attentional processing is compromised. In other words, a lack of prefrontal lobe activation to compensate for emotional dysregulation in mania may result in subcortical activation/cognitive decrements over time.
It is unlikely that these findings are substantially influenced by confounding factors. The study groups were well matched on most demographic and clinical variables, with the exception of years of education and IQ scores. Despite these differences, on average, members of both groups had college-level educations and were of average intelligence. Based on previous research and published normative data for clinical CPTs, we expect few performance differences in individuals within these education and IQ ranges44
. Additionally, these factors were statistically controlled in the primary cognitive and imaging analyses. With respect to performance, even at vigilance period 3 the BP group performed well above chance, indicating that they continued to give good effort throughout the task. Moreover, although the HC group outperformed the BP group on the proportion of CPT-IP correct rejections, the null Time x Group interaction is a good indication that the two groups processed correct rejections in similar fashion. To be conservative, as this was the performance variable of interest in the primary ROI analyses, correct rejections were included (along with education and IQ) as covariates.
As with any clinical study, a number of limitations should be considered when interpreting the results. Foremost, the results of this study are somewhat limited by variations in the medication status of the BP group. A direct comparison indicated that activation was reduced in medicated relative to unmedicated patients in both right IFG and midline cerebellum. The relative lack of activation in IFG, in particular, in medicated patients may help explain why this region was not implicated in any of the primary imaging analyses as expected. However, in that IFG and cerebellar effects were null in the primary analysis, it is unlikely that medication status influenced the obtained results directly. The effect of medication on distal portions of the anterior-limbic network through direct IFG and cerebellar activity, though less likely, cannot be ruled out. This sub-analysis should be interpreted cautiously, however, due to the small sample of unmedicated patient participants (n=10) and the possibility that unmedicated patients are simply healthier. Future studies using larger samples and functional connectivity approaches could help flesh out medication effects.
A second limitation is the inclusion of participants with past alcohol or substance abuse which might also have influenced the results and be considered a limitation. However, patient participants with a history of past alcohol abuse did not have outlying values (i.e., scores > 2 SD from the group mean or scores < the 5th
or > the 95th
percentile) on any cognitive or activation measure. Previous research demonstrates that the three month time frame we used to define abuse remission is sufficient for structural abnormalities in primary alcoholism to return to healthy volumetric values45
. Third, unmedicated patients were required to have not taken psychotropic medications for at least 48 hours, but certain antipsychotic and antidepressant agents may require longer washout periods. Although the actual time frame was not determined due to the difficulty in getting accurate estimates, it should be noted that participants in the unmedicated group generally consisted of inpatients admitted for an acute exacerbation of mania due to medication non-adherence. Therefore, it is likely that the medication-free period would have been greater than two days. Finally, anxiety disorders were not assessed on the SCID-I/P. Although all participants with bipolar disorder in the present study were hospitalized and treated for primary mood symptoms, the possible influence of concurrent anxiety disorders on the findings, and the lifetime incidence of anxiety disorders, is unknown.
The present study suggests that sustained attention deficits characterizing manic patients with bipolar disorder have measurable neurophysiological correlates. The findings indicate that amygdala overactivation characterizes manic bipolar disorder, even on the emotionally neutral CPT-IP version, and is independent of time in left amygdala. Superior frontal gyrus activation, on the other hand, occurs similarly in both bipolar and healthy individuals, and appears to represent the primary neurophysiological response to sustained attention in general. The greatest differential impact of sustained attention processing is on subcortical structures, including striatal structures (caudate and putamen) and thalamus. The extent to which these abnormalities represent a failure of cortical constraint on amygdala overactivation requires investigations of patients with bipolar disorder across mood states. During mania, the amygdala has been shown to be overactivated20,22
, while during euthymia VLPFC activation is enhanced 21
, possibly as a compensatory process to help restrain emotional input and sustain attention through the cortical control of effort. Although we are undertaking projects including both manic and euthymic groups to address this issue, the present findings provide the first evidence that acute mania involves a sustained attention decrement associated with differential neurophysiological processing over time.