Neural Reorganization Due to Neonatal Amygdala Lesions in the Rhesus Monkey: Changes in Morphology and Network Structure

Abstract

It is generally believed that neural damage that occurs early in development is associated with greater adaptive capacity relative to similar damage in an older individual. However, few studies have surveyed whole brain changes following early focal damage. In this report, we employed multimodal magnetic resonance imaging analyses of adult rhesus macaque monkeys who had previously undergone bilateral, neurotoxic lesions of the amygdala at about 2 weeks of age. A deformation-based morphometric approach demonstrated reduction of the volumes of the anterior temporal lobe, anterior commissure, basal ganglia, and pulvinar in animals with early amygdala lesions compared to controls. In contrast, animals with early amygdala lesions had an enlarged cingulate cortex, medial superior frontal gyrus, and medial parietal cortex. Diffusion-weighted imaging tractography and network analysis were also used to compare connectivity patterns and higher-level measures of communication across the brain. Using the communicability metric, which integrates direct and indirect paths between regions, lesioned animals showed extensive degradation of network integrity in the temporal and orbitofrontal cortices. This work demonstrates both degenerative as well as progressive large-scale neural changes following long-term recovery from neonatal focal brain damage.

Keywords: brain damage; connectome; morphometry; plasticity; tractography.

Figure 1.

Image processing and analysis overview. Brain morphometry and connectivity were analyzed using a step-wise processing pipeline: (1) T1-weighted images were deformably registered to a study-specific average template, facilitating analysis of targeted regional morphometry and whole-brain voxelwise morphometry. (2) Template image was segmented into subcortical gray, cortical gray, white matter, and CSF. (3) Template segmentation was propagated back to subject's native space. (4) Cortical parcellation (the RM) and manual subcortical tracings in template space were propagated into subject native space, and intersected with the native segmentation. (5) Diffusion-weighted image was used to perform probabilistic tractography, seeding streamlines from all white matter voxels. (6) Streamlines terminating in gray matter ROIs were used to construct a weighted connectivity matrix for each subject. Three different graph theoretical measures related to mutual communication (direct connection strength, path length, and communicability) were computed for each edge (i.e., each ROI pair).

Figure 2.

MRI and histological sections through the amygdala. Matching coronal MRI and Nissl-stained sections taken at 6 levels (arranged from rostral A to caudal F) for control monkey CM1 (A–F) and lesioned case LM3 (G–L), ordered from rostral (A and G) to caudal (F,L). The amygdala (A) is indicated in the control animal as is the entorhinal cortex (EC). The anterior commissure (ac) is prominent in the control animal but markedly shrunken in the lesioned animal. Note that the temporal horn of the lateral ventricle (V) is much expanded in the lesioned animal. Scale bar, 5 mm. Abbreviations: A, amygdala; EC, entorhinal cortex; V, ventricle; ac, anterior commissure.

Figure 3.

Extraneous damage. Histological sections from case LM3 showing the extent of extra-amygdala damage in the medial temporal lobe. The amount of extraneous damage was judged to be average in this animal and thus provides a summary of consistent damage across the experimental group. Sections are arranged from rostral (A,B) to caudal (C,D). At the level of the amygdala, direct lesional damage was seen in the ventral claustrum (CLv), the fundus of the superior temporal sulcus (fsts), the fundus of the anterior middle temporal sulcus (famts) and the fundus of the rhinal sulcus (frs). There was generally some damage to anterior levels of the hippocampus. In this case, cell loss is apparent in the dentate gyrus and in the CA1 field of the hippocampus (asterisks). Abbreviations: CLv, ventral claustrum; A, amygdala; CA1, CA1 field of the hippocampus; DG, dentate gyrus; EC, entorhinal cortex; famts, fundus of the anterior middle temporal sulcus; frs, fundus of the rhinal sulcus; fsts, fundus of the superior temporal sulcus; rs, rhinal sulcus; amts, anterior middle temporal sulcus; V, ventricle; ac, anterior commissure; sts, superior temporal sulcus.

Figure 4.

Profiles of regional morphometry in lesioned versus control animals. (A) Average whole brain tissue volumes (cortical GM, WM, and subcortical GM) and intracranial volumes (ICV), per group. Error bars represent standard deviation. Lesioned animals had smaller brains on average as evidenced by lower average ICV and lower volumes in each tissue class (N.S., though see text regarding effect sizes). (B) Regional between-group comparisons of jacobian determinants (values averaged across hemispheres) obtained from the nonlinear deformation procedure after accounting for total brain size. Jacobian determinants signify the ratio of subject vs template volume. Thus, lower values indicate less volume. Overall, 16 nonoverlapping ROIs are plotted, separated into 5 distinct subsystems and ordered from left-to-right within each subsystem according to known strength of amygdala connectivity (see Methods). Volumetric profiles were distinct across groups (P < 0.001 via rm-ANOVA). Rostral inferior temporal gyrus and perirhinal cortex were significantly smaller in lesioned animals versus controls, while middle cingulate cortex was significantly larger in lesioned animals versus controls. **P < 0.005, *P < 0.05. Abbreviations: ACC, anterior cingulate cortex; MCC, middle cingulate cortex; PCC, posterior cingulate cortex; Ia, agranular insular cortex; Ig_Id, granular/dysgranular insular cortex; med_orbit, medial orbitofrontal cortex; lat_orbit, lateral orbitofrontal cortex; lat_lat_orbit, lateral orbitofrontal/ventrolateral frontal cortex; ITG_ros, rostral inferior temporal gyrus; temp_pole, temporal polar cortex; STG_ros, rostral superior temporal gyrus; ITG_caud, caudal inferior temporal gyrus; STG_caud, caudal superior temporal gyrus; prh, perirhinal cortex; ec, entorhinal cortex; ph, parahippocampal cortex.

Figure 5.

Voxelwise morphometry demonstrates precise areas of lesion-induced shrinkage and cortical expansion. (A) The 4-year dataset. Significantly reduced volume was observed around the amygdala and anatomically connected structures, including perirhinal, entorhinal, rostral inferior temporal, and superior temporal cortices as well as subcortical structures such as the ventral putamen, claustrum, anterior hippocampus, and pulvinar. On the other hand, most of the rostrocaudal extent of the cingulate cortex was expanded, as well as parts of the cingulate bundle, medial superior frontal gyrus, and medial parietal cortices. (B) The 12-year dataset. Results were highly consistent with the 4-year dataset. Shrinkage of the anterior commissure was observable, perhaps due to higher spatial resolution and contrast-to-noise ratio. Cortical expansion was observable in the same areas shown in the 4-year data. See also surface projections of these data in Figure 7.

Figure 6.

Group differences in network structure using 3 graph metrics. (A–F) Each column shows results of the Network-Based Statistic (NBS) method for comparing group differences in network structure between controls vs. lesioned animals. Group differences were assessed using 3 different graph theoretic measures of mutual communication: direct connectivity strength (A, B), inverse path length (C, D), and communicability (E, F). In the top row, for each graph metric (A, C, and E), the size of the largest cluster of edges demonstrating Control > Lesion effects is plotted. The cluster size indicates the number of contiguous edges exceeding a givenT-statistic threshold. The next row (B, D, and F) shows the FWE-corrected P-value for the corresponding cluster at each T. Dashed lines are displayed at P = 0.05 and 0.01. The legend indicates different matrix densities imposed on the raw connectivity matrix, from which all graph metrics were computed. Gray bars in (C) and (E) indicate the range of cluster size at which we obtained FWE-corrected P < 0.05 for all matrix densities. Note that the path length measure is identical across matrix sparsities since the addition of very small weights do not alter the shortest paths. (G–I) In each column, the largest cluster of edges showing group differences are shown. All effects shown represent Control > Lesion effects. Results are displayed using the raw matrix density of 35% and edge-wise T-stat threshold of 2.8 (P-values shown were obtained with the NBS). Node sizes indicate the sum of suprathreshold effects across the adjoining edges. Each column defines the edge weight differently: direct connectivity strength (G), inverse path length (H), and communicability (I).

Figure 7.

Relationship of morphometric effects with previous findings of structural and functional effects of amygdala pathology. (A) Morphometric effects of neonatal amygdala lesions in 12-year old rhesus macaques. Same data as presented in Figure 4B, projected onto the cortical surface. (B) Effects of transient amygdala inactivation on regional functional connectivity in rhesus macaques (as previously published). Functional connectivity was defined as the strength of a region's spontaneously correlated activity with other brain areas. Regions that exhibited transiently increased functional connectivity, as a result of amygdala inactivation, encompassed the ACC and primary motor and somatosensory cortex. Decreased functional connectivity was seen in the amygdala, medial orbitofrontal cortices, and especially temporal polar and ventral temporal cortices. Visualization adapted from data presented in (Grayson et al. 2016). (C) Altered cortical thickness in adult human patients with developmental amygdala damage due to Urbach-Wiethe, a rare genetic disease. Results were assessed in 2 patients relative to a cohort of age-matched controls. Significantly increased cortical thickness was seen bilaterally in the ventromedial prefrontal cortex and cingulate cortex, especially the ACC. Reduced thickness was seen bilaterally around the amygdala. Adapted from (Boes et al. 2012). Coloring scheme was adjusted to match that used in (A) and (B).