Neuroanatomical connections and specific regional vulnerability in Alzheimer's disease

Abstract

PATHOLOGICAL changes associated with Alzheimer's disease (AD) occur in a consistent distribution, severely affecting certain neuronal populations in limbic areas and association cortices. De Lacoste and White (9) argue that the basis of this pattern lies in the connectional relationship among affected neurons. Indeed, in many instances, neurofibrillary tangles occur in projection neurons and senile plaques are found in termination zones. However, there are also counter-examples to this postulate. We suggest that, although disruption of these neural system connections probably underlies cognitive impairment in AD, the basis of selective neuronal vulnerability reflects an interaction between connectivity and specific molecular phenotypes in different brain regions. Whether the development of neurpathological changes associated with AD follows a predictable pattern is a critical issue in understanding the progression of the disease. The model of cortical disconnection and vulnerability due to cortical connectivity, described here by de Lacoste and White (9), proposes that pathological changes in certain brain regions can spread to other regions by well-defined cortical connections. The nature of this spread could be through the transfer of toxins or viruses, or through the disruption of interneuronal connections that normally help maintain neuronal viability. Furthermore, de Lacoste and White identify the interconnected regions of the limbic system and the temporo-occipital cortices as structures suited to test whether the AD-related changes spread from one region of the brain to another. It is important to divide the disconnection~connectivity model of AD into two components. The first is that disruption of connections of neural systems accounts for the clinical manifestations of AD. There are several pieces of data that support this assertion. First, the memory deficit of AD patients could be explained by the disruption of the connections of the hippocampus, amygdala, and entorhinal cortex by the presence of senile plaques (SP's) and neurofibrillary tangles (NFT's). These changes in the medial temporal lobe are the first alterations found in the brains of patients with AD (2,19,20,33) and are consistent with the primacy of memory loss in patients with AD. Second, the distribution and degree of neuropathological changes correlates, at least in some studies, with measures of dementia (2,27,37). Third, the appearance of NFT's and SP's in the visual regions of the occipital lobe correlates with deficits in visuospatial skills in the small subpopulation of AD patients with Balint's syndrome (14). Finally, it seems likely that the correlation between degree of cognitive impairment and the loss of synapses seen in AD reflects a loss of cortico-cortical projections (37). Thus, it appears that disruption of cortical connections can account for the type and severity of the clinical symptoms in AD. The second component of the disconnection/connectivity model is that the pattern of NFT's and SP's in the brains of patients with AD can be explained by the spread of pathological changes through interneuronal connections. One example that supports this concept is the analysis of hippocampal formation damage in AD. NFT's are observed in layer II neurons of the entorhinal cortex whose axons project to the molecular layer of dentate gyrus, an area frequently involved by SP's ( 16,17). SP's also accumulate in specific nuclei in the amygdala that receive hippocampal projections (19,24). NFT's accumulate in high-order association cortices ( 14,25,31), whose projections to other cortices terminate in regions affected with SP's. These observations are consistent with the hypothesis that SP's develop in the terminal zone of neurons that contain NFT's. Alternately, this data could be interpreted to imply that NFIs form in cells projecting to SP's. However, one can also find instances of areas severely affected by NFIs whose projection zone is not frequently affected by SP's. For example, subicular/CA 1 hippocampal neurons develop NF'Is. These neurons project strongly to layer IV of the entorhinal cortex which contains NFT's but little or no amyloid deposition (19). Other apparent discrepancies are found in an analysis of the visual system as described below. Thus, while it may be possible to predict which regions of the brain are affected at different stages of AD, it does not seem possible to predict the pattern of involvement of different brain regions based only on known anatomical connections.