Diverse pathophysiological processes converge on network disruption in mania

Abstract

Background: Neuroimaging of psychiatric disease is challenged by the difficulty of establishing the causal role of neuroimaging abnormalities. Lesions that cause mania present a unique opportunity to understand how brain network disruption may cause mania in both lesions and in bipolar disorder.

Methods: A literature search revealed 23 case reports with imaged lesions that caused mania in patients without history of bipolar disorder. We traced these lesions and examined resting-state functional Magnetic Resonance Imaging (rsfMRI) connectivity to these lesions and control lesions to find networks that would be disrupted specifically by mania-causing lesions. The results were then used as regions-of-interest to examine rsfMRI connectivity in patients with bipolar disorder (n = 16) who underwent imaging longitudinally across states of both mania and euthymia alongside a cohort of healthy participants scanned longitudinally. We then sought to replicate these results in independent cohorts of manic (n = 26) and euthymic (n = 21) participants with bipolar disorder.

Results: Mania-inducing lesions overlap significantly in network connectivity. Mania-causing lesions selectively disrupt networks that include orbitofrontal cortex, dorsolateral prefrontal cortex, and temporal lobes. In bipolar disorder, the manic state was reflected in strong, significant, and specific disruption in network communication between these regions and regions implicated in bipolar pathophysiology: the amygdala and ventro-lateral prefrontal cortex.

Limitations: There was heterogeneity in the clinical characterization of mania causing lesions.

Conclusions: Lesions causing mania demonstrate shared and specific network disruptions. These disruptions are also observed in bipolar mania and suggest a convergence of multiple disorders on shared circuit dysfunction to cause mania.

Keywords: Bipolar; Disorder; Lesion; Mania; Network; fMRI.

Figure 1:. Tracings of Lesions that Induced Mania

Lesion location of 23 cases of lesion-induced mania, traced onto a standard template brain. Horizontal sections are oriented with the right side of the brain on the reade’s left.

Figure 2:. Network Mapping for Mania-inducing Lesions

23 mania-causing lesions were traced from published images in case reports onto a template brain (column A). Each lesion tracing was used as a ROI for mapping functional connectivity to this region in healthy control subjects (B), allowing visualization of the network disrupted by the lesion. Regions of positively correlated activity to the ROI are shown in hot colors and negatively correlated activity in cold colors. These lesion network maps were overlaid to identify convergent areas of network connectivity among mania-inducing lesions (C). Left Color Bar: T-statistic. Right Color Bar: Number of lesion maps converging on a given voxel.

Figure 3:. Anatomical Specificity of Lesion-induced Mania

Network connectivity to mania-causing lesions is compared to network connectivity in a set of control stroke lesions. A voxelwise Liebermeister test reveals regions significantly (p-FDR <.05) more functionally connected to mania-causing lesions than controls lesions. Regions of positively correlated activity to the ROI are shown in hot colors and negatively correlated activity in cold colors. Three regions of significant differences in connectivity were observed: The Orbitofrontal Cortex (OFC), the bilateral temporal lobe, and the bilateral Dorso-Lateral Pre-Frontal Cortex (DLPFC, Brodmann Area 46). The OFC and temporal lobes were intrinsically correlated with areas of mania-causing lesions. BA46 was intrinsically anti-correlated with mania-causing lesions. None of the control lesions were connected to these areas. Color Bar: Z-stat.

Figure 4:. Lesion-induced and Bipolar Mania Converge On Shared Network Dysconnectivity

We sought to determine if the networks identified in the lesion mapping analysis of mania were also disrupted in bipolar mania. Using the regions identified in lesion-induced mania as ROIs, we compared functional connectivity to these areas in participants with Bipolar Disorder type I imaged in a manic state and then in a euthymic state. A) Using the temporal pole ROIs from lesion network mapping (inset), we observe disconnectivity in mania to the bilateral Ventro-Lateral Pre-Frontal Cortex (VLPFC) (Right: k=47, peak voxel T-stat 5.89 p<.001, x+57 y+27 z0; Left: k=32, peak voxel T-stat 5.39 p<.001, x-54, y+21, z-3). B) The BA46 ROIs from the lesion mapping analysis (inset) demonstrate increased connectivity in mania to the left amygdala (k=29, peak voxel T-stat 4.74, p<.001, x-21 y0 z-24). Here increased connectivity indicates dis-connectivity between normally anti-correlated structures. BA46 also demonstrated increased connectivity to the dorso-medial Pre-Frontal Cortex (dmPFC) (k=32, peak voxel T-stat 4.97, p<.001, x-12, y+9, z+36). A set of healthy comparison participants imaged longitudinally across these time points did not demonstrate any significant change in connectivity in these regions. Images are thresholded voxelwise p<.001, cluster-extent p<.016. Color Bar: T-stat.

Figure 5:. Replication of Temporal-VLPFC Disconnectivity in Bipolar Mania Compared to Euthymia

We sought to determine if the networks identified in the lesion mapping analysis of mania also disrupted in bipolar mania. Using the regions identified in lesion-induced mania as ROIs, we compared functional connectivity to these areas in participants with Bipolar Disorder type I imaged in a manic state and then in a euthymic state. A) Using the temporal pole ROIs from lesion network mapping (inset), and examining the VLPFC regions identified in the longitudinal analysis (Figure 4), we again observe disconnectivity in mania to the bilateral Ventro-Lateral Pre-Frontal Cortex (VLPFC) (Right: k=52, peak voxel T-stat 3.78, p<.001, x45 y30 z-9; Left: k=23, peak voxel T-stat 3.33 p=.001, x-51, y27, z-9). While the VLPFC of both hemispheres showed reduced connectivity to the temporal lobe ROIs in mania, only the right VLPFC met extent significance (k=33 for p<.025 corrected for multiple comparisons). Images are thresholded voxelwise p<.05. Color Bar: T-stat.

Figure 6:. A Model of The Network Basis of Mania

We propose that multiple causes of mania converge on the same networks and that it is disruption of distributed brain networks that mediate the link between pathophysiology and behavioral phenotype. Specifically, this network can be disrupted by either acute lesioning of critical nodes of this network (e.g. temporal lobes, OFC) or may be due to a combination of genetic factors and environmental stressors that give rise to manic episodes in bipolar disorder. In either scenario, the disruption of communication across polysynaptic networks causes the manic phenotype. Within these networks, longitudinal imaging data supports disconnectivity between (A) Temporal poles and VLPFC and (B) Between amygdala and DLPFC (BA46) as causes of mania in bipolar disorder. Arrows here are not meant to denote a monosynaptic pathway.