Effects of neonatal amygdala or hippocampus lesions on resting brain metabolism in the macaque monkey: a microPET imaging study

Abstract

Longitudinal analysis of animals with neonatal brain lesions enables the evaluation of behavioral changes during multiple stages of development. Interpretation of such changes, however, carries the caveat that permanent neural injury also yields morphological and neurochemical reorganization elsewhere in the brain that may lead either to functional compensation or to exacerbation of behavioral alterations. We have measured the long-term effects of selective neonatal brain damage on resting cerebral glucose metabolism in nonhuman primates. Sixteen rhesus monkeys (Macaca mulatta) received neurotoxic lesions of either the amygdala (n=8) or hippocampus (n=8) when they were two weeks old. Four years later, these animals, along with age- and experience-matched sham-operated control animals (n=8), were studied with high-resolution positron emission tomography (microPET) and 2-deoxy-2[(18)F]fluoro-d-glucose ([(18)F]FDG) to detect areas of altered metabolism. The groups were compared using an anatomically-based region of interest analysis. Relative to controls, amygdala-lesioned animals displayed hypometabolism in three frontal lobe regions, as well as in the neostriatum and hippocampus. Hypermetabolism was also evident in the cerebellum of amygdala-lesioned animals. Hippocampal-lesioned animals only showed hypometabolism in the retrosplenial cortex. These results indicate that neonatal amygdala and hippocampus lesions induce very different patterns of long-lasting metabolic changes in distant brain regions. These observations raise the possibility that behavioral alterations in animals with neonatal lesions may be due to the intended damage, to consequent brain reorganization or to a combination of both factors.

Figure 1

Six coronal images, arranged from rostral (top) to caudal (bottom), are shown from the rhesus monkey MRI atlas created for MRI/microPET image registration (left column). Six registered microPET images representing the average glucose metabolism in the eight sham-operated control animals (CON) are also shown to demonstrate microPET image resolution and multi-modal spatial registration accuracy. On these, and all subsequent coronal MRI and microPET images, the observer's left is the animal's left. Numbers to the left of each row represent the distance (in mm) of each image from the interaural plane. Colors indicating high, moderate and low glucose metabolism are indicated by the color bar.

Figure 2

Four coronal images, arranged from rostral (top) to caudal (bottom), are shown from the normalized rhesus monkey MRI atlas (left column), and averaged MRIs of all hippocampal-lesioned (center column) and amygdala-lesioned (right column) animals to demonstrate common areas of atrophied tissue (dark bands at arrow tips) or enlarged ventricles (dark areas below asterisks) in the lesioned animals. Numbers to the left of each row represent the distance (in mm) of each image from the interaural plane. The entire brain surface was traced in white on the four images from the rhesus monkey MRI atlas, and these boundaries were transferred directly onto registered images from the two lesioned groups. For the hippocampal-lesioned group, atrophied or shrunken tissue typically occurred around the parahippocampal cortex (level +4) and the temporal horns of the lateral ventricle are enlarged (levels +10 and +4). For the amygdala-lesioned animals, shrunken tissue typically occurred in the temporal pole, area TE, perirhinal cortex and entorhinal cortex (levels +23 and +17), and enlarged ventricles are also evident at level +17.

Figure 3

Coronal images, arranged from rostral (top, left) to caudal (bottom, right), from the rhesus monkey MRI atlas are shown with the twenty ROIs used in this study. These sample images are spaced 5 mm apart, and their distance (in mm) from the interaural plane is given in the upper-right corner of each image. The following ROIs were drawn: A) frontal pole, B) ventrolateral prefrontal cortex, C) dorsolateral prefrontal cortex, D) anterior cingulate cortex, E) orbital and ventromedial frontal cortex, F) temporal pole, G) putamen, H) caudate nucleus, I) nucleus accumbens, J) substantia innominata, K) globus pallidus, L) area TE, M) perirhinal cortex, N) amygdala, O) entorhinal cortex, P) hippocampus, Q) parahippocampal cortex, R) area TEO, S) retrosplenial cortex and T) cerebellum.

Figure 4

Relative metabolism in the cerebellar ROI for each group. In this and all subsequent bar graphs, relative metabolism is displayed in terms of the percent difference from whole brain mean RI count (5000), and was calculated as follows: % Difference = ((mean ROI RI count − 5000)/5000) × 100. Vertical lines associated with each bar represent the standard error of the mean. ** p < .01.

Figure 5

Two coronal and one sagittal MRI/microPET fusion images with the cerebellar ROI outlined. Numbers to the left of coronal images indicate distance (in mm) from the interaural plane, whereas the number to the left of the sagittal image indicates distance (in mm) from the midsagittal plane (positive numbers indicate right hemisphere). The levels of the two coronal images are also indicated in red on the sagittal image. The orange overlay indicates areas where average RI count was more than two standard deviations greater in amygdala-lesioned animals relative to controls.

Figure 6

The percent difference from mean whole brain metabolism in the five frontal lobe ROIs for each experimental group. For each ROI, the percent difference shown is an average across the left and right hemispheres. Group differences were detected for the orbital and ventromedial frontal cortex, anterior cingulate cortex and the dorsolateral prefrontal cortex, but not the frontal pole or ventrolateral prefrontal cortex. Vertical lines associated with each bar represent the standard error of the mean. * p ≤ .05.

Figure 7

Two coronal and one sagittal MRI/microPET fusion images are shown through the frontal lobes with the dorsolateral prefrontal cortex (white), anterior cingulate cortex (blue) and orbital and ventromedial frontal cortex (yellow) ROIs outlined. Numbers to the left of each image indicate distance (in mm) from the interaural plane for coronal images or from the midline for sagittal images (negative numbers indicate left hemisphere). The levels of the two coronal images are also indicated in red on the sagittal image. The orange overlay indicates areas where average RI count was more than two standard deviations greater in control animals than amygdala-lesioned animals.

Figure 8

The percent difference from mean whole brain metabolism in the five subcortical ROIs (averaged across hemispheres) for each experimental group. Significant group differences were detected for the caudate nucleus and putamen, but not for the globus pallidus, nucleus accumbens or substantia innominata. Vertical lines associated with each bar represent the standard error of the mean. * p ≤ .05.

Figure 9

Two coronal MRI/microPET fusion images are shown with the caudate nucleus (white) and putamen (blue) ROIs outlined. Numbers to the left of each image indicate distance (in mm) from the interaural plane. The orange overlay indicates areas where average RI count was more than two standard deviations greater in control animals than amygdala-lesioned animals.

Figure 10

The percent difference from mean whole brain metabolism in the nine temporal and posterior cingulate ROIs (left and right hemispheres averaged) for each experimental group. Significant group differences were only detected for the hippocampus and retrosplenial cortex. Vertical lines associated with each bar represent the standard error of the mean. * p ≤ .05, ** p < .01.

Figure 11

Sagittal MRI/microPET fusion images showing the right hippocampus (left panel) and left retrosplenial cortex ROIs (right panel). Numbers to the left of these sagittal images indicate distance (in millimeters) from the midline (positive numbers indicate right hemisphere, negative numbers indicates left hemisphere). The orange overlay indicates areas where average RI count differed by more than two standard deviations between groups as indicated.