Connectivity Changes from Head Injury

<p>From: <a href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414">Resting-State Functional Magnetic Resonance Imaging Activity and Connectivity and Cognitive Outcome in Traumatic Brain Injury | Radiology | JAMA Neurology | JAMA Network</a></p>

<p class="para">Cognitive deficits after traumatic brain injury (TBI) are a major cause of daily life disability.<sup><a class="ref-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036r1" data-tab-toggle=".tab-nav-references">1</a></sup>Persistent disabling sequelae are caused by the structural brain damage that occurs not only in the early stage but also after a period of apparent recovery.<sup><a class="ref-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036r2" data-tab-toggle=".tab-nav-references">2</a></sup>Traumatic axonal injury is associated with the most severely impaired outcomes after TBI. Neuropathological studies have shown that TBI affects structural brain networks progressively, from focal axon alteration to delayed axonal disconnection.<sup><a class="ref-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036r3" data-tab-toggle=".tab-nav-references">3</a></sup>Functional magnetic resonance imaging (fMRI) studies suggest that white matter damage alters structural connectivity, which in turn can affect functional connectivity, and both contribute to cognitive dysfunctions in TBI.<sup><a class="ref-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036r4" data-tab-toggle=".tab-nav-references">4</a></sup><sup>-<a class="ref-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036r4" data-tab-toggle=".tab-nav-references">7</a></sup></p>

<p class="para">Resting-state fMRI studies have recently emerged as a useful tool for investigating brain functional connectivity after severe TBI<sup><a class="ref-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036r4" data-tab-toggle=".tab-nav-references">4</a></sup><sup>,<a class="ref-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036r8" data-tab-toggle=".tab-nav-references">8</a></sup><sup>-<a class="ref-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036r8" data-tab-toggle=".tab-nav-references">11</a></sup> and in subjects with mild TBI examined during the early stages.<sup><a class="ref-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036r12" data-tab-toggle=".tab-nav-references">12</a></sup><sup>,<a class="ref-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036r13" data-tab-toggle=".tab-nav-references">13</a></sup> These studies provide information about brain activity and connectivity in the absence of task performance, a condition that allows researchers to investigate patient populations with broader ranges of injury severity, because no specific cognitive ability is required. Advanced neuroimaging processing tools provide information about brain activity and functional connectivity during resting-state fMRI. Low signal fluctuations that occur during rest have been shown to reflect strong connectivity between functionally related brain regions.<sup><a class="ref-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036r14" data-tab-toggle=".tab-nav-references">14</a></sup><sup>,<a class="ref-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036r15" data-tab-toggle=".tab-nav-references">15</a></sup></p>

<p class="para"><a name="q1"></a><span class="quizSection">Although different functional connectivity networks can be identified during the resting state,<sup><a class="ref-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036r16" data-tab-toggle=".tab-nav-references">16</a></sup> the most widely studied is the default mode network (DMN), which is reported to be affected in a broad range of brain disorders and is commonly related to cognitive processes.<sup><a class="ref-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036r17" data-tab-toggle=".tab-nav-references">17</a></sup> The spatial pattern of the DMN includes the ventromedial prefrontal cortex, the posterior cingulate cortex, the lateral parietal cortex, and the precuneus.</span> For the main nodes of the DMN, their functional connectivity is supported by an underlying structure of white matter pathways,<sup><a class="ref-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036r18" data-tab-toggle=".tab-nav-references">18</a></sup>with the cingulum as the key tract that interconnects the anterior and posterior core regions of the DMN.<sup><a class="ref-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036r19" data-tab-toggle=".tab-nav-references">19</a></sup></p>

<p class="para"><a name="q2"></a><span class="quizSection">In addition to disrupted functional connectivity, studies of white matter integrity using diffusion tensor imaging<sup><a class="ref-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036r20" data-tab-toggle=".tab-nav-references">20</a></sup><sup>-<a class="ref-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036r20" data-tab-toggle=".tab-nav-references">23</a></sup> have shown that structural connectivity is also greatly affected after TBI.</span> In a previous study,<sup><a class="ref-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036r5" data-tab-toggle=".tab-nav-references">5</a></sup> we found that disrupted structural connectivity after traumatic axonal injury was responsible for altered functional connectivity during working memory performance. In the current research, we used new advances in resting state–related fMRI methods to determine whether the amplitude of spontaneous low-frequency fluctuations in brain activity and DMN connectivity may be sensitive biomarkers for cognitive dysfunction after TBI. To this end, we measured the amplitude of resting-state blood oxygen level–dependent signal fluctuations in the whole brain and the DMN, studying global and network-based functional connectivity, as well as structural connectivity of the main fasciculi within the DMN.</p>

<div class="heading-text ">Amplitude of Resting-State Fluctuations</div>

</div>

<p class="para">Compared with controls, patients had greater ALFF in several brain areas (corrected <em>P</em> < .05), including the frontal pole, superior frontal gyrus, middle frontal gyrus, paracingulate gyrus, and superior parietal lobe (see <a class="figure-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036f2" data-tab-toggle=".tab-nav-figure-table">Figure 2</a> and <a class="table-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036t2" data-tab-toggle=".tab-nav-figure-table">Table 2</a> for MNI coordinates and cluster size). Because the map of increased ALFF in the middle frontal areas showed a large overlap with the frontal node of the DMN, we analyzed the brain connectivity of this network.</p>

<div class="heading-text ">Independent Component Analysis</div>

</div>

<p class="para">Using ICA with temporal concatenation, we obtained a set of 37 independent components and identified the main resting-state networks (eResults and eFigure 1 in <a class="supplement-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#note-NOI130036-1" data-tab-toggle=".tab-nav-supplemental">Supplement</a>). Within this set of networks, we selected the DMN for further analysis.</p>

<div class="heading-text ">DMN ConnectivityGroup</div>

</div>

<div class="heading-text ">Comparisons of the ICA-Based DMN</div>

</div>

<p class="para">Using the dual regression approach, patients with TBI had greater functional connectivity than controls within the DMN in certain areas of the frontal lobe (<a class="figure-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036f3" data-tab-toggle=".tab-nav-figure-table">Figure 3</a>). These regions included the frontal pole, the anterior cingulate and paracingulate gyrus, the superior frontal gyrus, and a small part of the precentral gyrus.</p>

<div class="heading-text ">Seed-Based Connectivity Analysis</div>

</div>

<p class="para">Using the ICA-based spatial map of the DMN, we created 4 spherical ROIs with which to compute seed-based connectivity maps in the MPFC, PPC, and left and right parietal cortex ROIs (see eResults and eFigure 2 in <a class="supplement-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#note-NOI130036-1" data-tab-toggle=".tab-nav-supplemental">Supplement</a> for exact ROI locations).</p>

<p class="para">With the MPFC ROI as a seed, we found increased functional connectivity of this region with other areas of the MPFC in patients compared with controls (corrected <em>P</em> < .05) (<a class="figure-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036f4" data-tab-toggle=".tab-nav-figure-table">Figure 4</a>). Moreover, with the left parietal cortex ROI, we also found a set of brain areas with increased functional connectivity in patients with TBI (corrected <em>P</em> < .05) (<a class="figure-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036f4" data-tab-toggle=".tab-nav-figure-table">Figure 4</a>). These areas were located in the left lateral occipital cortex, angular gyrus, supramarginal gyrus, temporo-occipital areas, middle and inferior temporal gyrus, occipital fusiform and lingual gyri, and precuneus. We found no differences in functional connectivity when using the remaining nodes as seeds.</p>

<div class="heading-text ">Structural Connectivity Measured With Diffusion Tensor Imaging</div>

</div>

<p class="para">The MPFC and the PPC ROIs were used to reconstruct the white matter pathway connecting the 2 regions. The probabilistic tractography map of each subject was used to create separate average maps for patients and controls. These maps indicate the probability of each voxel being part of a white matter pathway connecting the 2 regions. The main tracts identified were the left and right bundles of the cingulum. Visual inspection of the average maps for each group showed that the size of the cingulum was reduced in patients.</p>

<p class="para">The average map for the control group was used as a mask to extract mean FA values within the cingulum for each subject. These values were significantly decreased in patients with TBI compared with controls (mean, 0.36 for patients vs 0.42 for controls; <em>t</em> = 5.9; <em>P</em> < .001) (<a class="figure-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036f5" data-tab-toggle=".tab-nav-figure-table">Figure 5</a>).</p>

<div class="heading-text ">Amplitude of Fluctuations and Cognitive Outcome</div>

</div>

<p class="para">In patients, the amplitude of resting-state fluctuations was positively correlated with cognitive performance (<em>r</em> = 0.48; <em>P</em> = .03); the greater the activation, the better the cognitive outcome. There was no significant correlation between these measures in controls (<em>r</em> = 0.35; <em>P</em> = .16).</p>

<p class="para">Within the patient group, functional connectivity scores for the frontal ROI were negatively correlated with the FA values of the cingulum tract (<em>r</em> = −0.45; <em>P</em> = .04).</p>

<div class="heading-text thm-col">Discussion</div>

</div>

<p class="para">The purpose of this study was to provide further insight into the ALFF and their connectivity in the resting state and the possible relationship of these findings with cognitive outcome after traumatic axonal injury. <a name="q3"></a><span class="quizSection">Our main finding was that higher ALFF at rest is associated with better cognitive outcome in patients with diffuse TBI. More specifically, these patients also had increased resting-state functional connectivity in regions surrounding the frontal node of the DMN.</span> Moreover, the increased frontal connectivity could be explained by damage to the cingulum, the key tract connecting the anterior and posterior brain areas of the DMN. These findings suggest that the loss of structural connectivity is compensated for by an increase in the functional connectivity of local circuits.</p>

<p class="para"><a name="q4"></a><span class="quizSection">Few studies to date have considered the ALFF at rest and its implications in terms of cognitive dysfunction. This measure has been found to be decreased in patients with mild cognitive impairment and Alzheimer disease.</span><sup><a class="ref-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036r31" data-tab-toggle=".tab-nav-references">31</a></sup> In neurodegenerative diseases, these decreases in amplitude probably reflect a loss of neurons that consecutively provokes connectivity deficits and disorganization or breakdown of brain networks. In our study, when comparing whole-brain ALFF in the resting state between groups, the TBI group showed increased amplitudes that were predominantly focused within frontal lobe regions. Moreover, the measures of higher amplitudes in these areas predicted a better general cognitive outcome, suggesting that functional and efficient brain reorganization occurred to compensate for acute brain damage and improve cognitive performance. This finding, together with changes in structural connectivity, may constitute an objective measure of long-term cognitive outcome after TBI.</p>

<p class="para">The specific analyses of the DMN in the resting state also showed increased connectivity in the frontal node of this network in patients with TBI compared with controls. Furthermore, connectivity from each core of DMN nodes to the whole brain revealed a widespread pattern of locally increased functional connectivity surrounding the medial frontal and left parietal nodes of the DMN. Increased functional connectivity has been found in other diseases involving white matter damage, such as multiple sclerosis; subjects with relapsing-remitting multiple sclerosis have compensatory increased connectivity in the posterior cingulate DMN node,<sup><a class="ref-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036r32" data-tab-toggle=".tab-nav-references">32</a></sup> and in patients with early-stage disease, distinct networks exhibit increases in functional connectivity despite large reductions in white matter integrity.<sup><a class="ref-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036r33" data-tab-toggle=".tab-nav-references">33</a></sup></p>

<p class="para">Previous studies investigating the DMN during the resting state in TBI have found alterations that reflect both decreases<sup><a class="ref-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036r11" data-tab-toggle=".tab-nav-references">11</a></sup> and increases in functional connectivity.<sup><a class="ref-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036r12" data-tab-toggle=".tab-nav-references">12</a></sup>The increases have been interpreted as compensatory or adaptive mechanisms because they were often positively correlated with cognitive outcome.<sup><a class="ref-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036r8" data-tab-toggle=".tab-nav-references">8</a></sup><sup>,<a class="ref-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036r34" data-tab-toggle=".tab-nav-references">34</a></sup> Another possible interpretation is that this increased connectivity, measured in voxels surrounding the DMN nodes, reflects a diffusion of the DMN nodes during recovery from TBI. This latter interpretation also may be supported by the fact that the white matter injury found in the cingulum in TBI precludes the functional inhibition of areas surrounding these nodes. We suggest that the increases found in frontal areas may reflect compensation because the amplitude of the fluctuations in these regions correlated with performance, but the increase in the spread of connectivity around the PPC may reflect a loss of directionality in the connectivity caused by the injury. On the other hand, Mayer et al<sup><a class="ref-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036r8" data-tab-toggle=".tab-nav-references">8</a></sup> described a pattern of decreased long-distance functional connectivity between the nodes of the DMN (ie, between anterior and posterior nodes).</p>

<p class="para">Although not directly examining the DMN, other authors have found alterations in resting-state connectivity in TBI in the form of reduced interhemispheric connectivity of the hippocampus and increased ipsilateral connectivity, and these were associated with better cognitive outcome.<sup><a class="ref-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036r6" data-tab-toggle=".tab-nav-references">6</a></sup> In addition, resting-state fMRI studies involving magnetoencephalographic recordings have also provided evidence of brain plasticity and network reorganization mechanisms, specifically increases or decreases in the extent of certain brain connections.<sup><a class="ref-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036r35" data-tab-toggle=".tab-nav-references">35</a></sup></p>

<p class="para">To find a possible explanation for the observed pattern of augmented functional connectivity in the anterior and posterior areas of the DMN, we performed tractography of the fibers connecting the core of the anterior and posterior regions of this network. This showed reductions within the fibers corresponding to the cingulum, and the FA of this bundle was correlated with the increased connectivity of the frontal node of the DMN. The cingulum is a long, medial associative bundle that runs within the cingulate gyrus all around the corpus callosum, connecting the medial frontal and parietal lobes.<sup><a class="ref-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036r36" data-tab-toggle=".tab-nav-references">36</a></sup> This finding suggests that damage to this fascicle leads to functional reorganization consisting of increased local connectivity surrounding the nodes of the network, probably resulting from decreased interconnectivity between the frontal and parietal nodes.</p>

<p class="para"><a name="q5"></a><span class="quizSection">Altogether, our results suggest that increased frontal functional activity at rest, measured as the ALFF, is associated with better global cognitive performance and that the altered structural connectivity between related brain regions can be compensated for by increased functional connectivity.</span> Two recent studies have accurately characterized the changes in the DMN connectivity at different time points. Hillary et al,<sup><a class="ref-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036r10" data-tab-toggle=".tab-nav-references">10</a></sup> examining resting-state DMN connectivity in a sample of patients in the acute stage with predominant focal lesions, found increases in the connectivity of this network during the first 6 months after injury. On the other hand, Arenivas et al<sup><a class="ref-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036r11" data-tab-toggle=".tab-nav-references">11</a></sup> examined DMN functional connectivity in a cohort of patients 6 to 11 months after TBI. Using 3 methodological approaches, they found decreased connectivity in the main nodes of the DMN. Although their sample characteristics were similar to our own (patients with white matter damage without significant contusions), our patients were studied a mean of 4 years after injury. Thus, our findings can be generalized only to patients in the late chronic stage and are not directly comparable to findings in the early chronic or postacute stages. However, whereas Arenivas et al<sup><a class="ref-link section-jump-link" href="https://jamanetwork.com/journals/jamaneurology/fullarticle/1688414#noi130036r11" data-tab-toggle=".tab-nav-references">11</a></sup> showed decreases in the core nodes of the DMN, we found diffuse increases in connectivity in areas surrounding DMN nodes. The apparent contradiction can be explained by the fact that Arenivas et al did not analyze the regions where we found increases because they studied connectivity within the nodes of the DMN. We suggest that the increased connectivity in areas surrounding DMN nodes, which we found in the late chronic stage, might represent the brain’s attempt to compensate functionally for weaker connectivity within DMN nodes caused by structural damage after traumatic axonal injury. Nevertheless, the decreased connectivity of the DMN in this study did not predict the clinical or cognitive deficits. Future longitudinal studies with several follow-up points could clarify the dynamics of cerebral reorganization after TBI.</p>