In 1906 the pathologist/psychiatrist Alois Alzheimer reported to a group of German psychiatrists the story of a 55-year-old woman who died after several years of progressive dementia, and in whose brain he had found both senile plaques and neurofibrillary tangles. Alzheimer had applied newly discovered silver stains to the tissue in order to better delineate these microscopic lesions. Plaques had been previously seen, presumably with aniline stains, but not in association with dementia, while cortical tangles had not been previously described. The head of Alzheimer’s department in Munich was Emil Kraepelin, who at the time was a dominant figure in European psychiatry, and who eagerly promoted the notion of psychiatric disease being based on organic brain changes. Kraepelin applied Alzheimer’s name to the newly described syndrome of dementiaplaques- tangles in the eighth edition of his important text Psychiatrie in 1910.

The two lesions, i.e., the intraneuronal neurofibrillary tangle and the neuropil plaque, are still regarded as the most important diagnostic markers and by most investigators as being the most important elements in the pathogenesis of dementia. The components of these lesions were not ascertained until the application of electron microscopy in the early 1960s, more than 50 years subsequent to Alzheimer’s first light microscopic description.

During that half century, however, several important findings were added to the picture and to our understanding. In 1911, Simchowicz reported the presence of granulovacuolar bodies in hippocampal pyramidal neurons and occasionally in neurons of the basal nucleus of Meynert. They are very rare elsewhere, and their significance is still unknown. In 1927, the Belgian Divry recognized that the more or less amorphous material in the core of the plaque is amyloid. This abnormal fibrillar protein, which is now known to be derived from a normal large precursor protein, is currently believed by many investigators to be the major pathogenetic factor. Scholtz recognized amyloid in the cortical and meningeal blood vessels in 1938, and it was the latter location which gave Glenner the opportunity to isolate and sequence the amyloid peptide some 45 years later. At the time of my first interest in Alzheimer’s disease in 1959, there was very little current epidemiologic, clinical, or investigative activity in the field. That most cases of senile dementia were due to Alzheimer’s disease was not recognized prior to Corsellis’ 1962 book. In 1959 Saul Korey, who was Chief of Neurology and a well-trained neurochemist at the Einstein College of Medicine, and Robert Terry decided to study the disease utilizing brain biopsy tissue, not because the clinical importance of the disorder was known to us, but rather because the problem seemed to be accessible to our technology. That was the case in the sense that the disease was fatal and thus the biopsy procedure could do relatively little harm even at worst. Second, the changes were diffuse so that the neurosurgeon could remove a small portion (less than 1 gram) from any “silent” region of the neocortex and could expect it to contain lesions. Third, it had distinctive histologic changes such that a histopathologic diagnosis could be made. A great neurosurgeon, Leo M. Davidoff, performed about a half dozen cortical biopsies at Einstein within the first few years, and none of these patients had any post-operative difficulties.

The tangles were easily found with the electron microscope, and were revealed to be made up curious, twisted fibers which Kidd, working simultaneously at Maida Vale Hospital with McMenemy in London, reported correctly as paired helical filaments (PHF), but which we in New York thought were twisted tubules. In any case, they did not appear to be identical to normal neurofilaments although the dimensions were about the same. Similarly, they appeared not to have the same electron density as neurotubules, which, in fact, were rather poorly preserved in the fixatives available at the time. PHF are now known to be made up largely of hyperphosphorylated tau protein, and this process of abnormal phosphorylation undoubtedly plays a major role in the overall pathogenesis of AD. The twisted structures of the tangle are not universally accepted as PHF. A flat twisted ribbon has been recently suggested by atomic force microscopy. The senile plaques were much more difficult to identify and delineate in those early days. We ultimately determined that at the center of these spherical lesions were more or less compact bundles of filaments representing amyloid fibers. Surrounding them were groups of unmyelinated neuronal processes, many of which were distended and contained PHF identical to those of the tangle. Many of these neurites terminated in presynaptic boutons which were also enlarged and contained many degenerating mitochondria and dense bodies, as well as clusters of synaptic vesicles. The synaptic gap and the postsynaptic terminal were usually normal. Most of the dystrophic neurites were, therefore, axonal. Astrocytic perikarya are often found on the periphery of the plaque with their filament-filled processes infiltrating the lesion. Microglia are also frequently present in and around the plaque. These cells are often deeply indented by finger-like projections of amyloid bundles. It was subsequently determined by others that in normal aging the neurites of the plaque, although dystrophic, did not contain PHF. In the normal elderly there may be large numbers of plaques, but few or no perikaryal tangles in the cortex; in the absence of tangles, PHF are not to be found in the plaques. These lesions in normal aging contain the same amyloid (β-protein) as do the plaques of the disease.

Most current investigators have not been able to find a significant correlation between the concentration of plaques and the severity of the dementia, in contrast to the earlier, very influential report of Blessed, Tomlinson and Roth. The amyloid burden does not seem to be the cause of dementia.

To add further confusion, some cases of clinically typical Alzheimer’s disease are found to have plaques in a concentration consistent with disease but without tangles in the neocortex. PHF are present, however, in entorhinal and hippocampal regions. About two thirds of these plaqueonly cases display Lewy bodies in the neocortex, and this latter situation has come to be called the “Lewy body variant”. However, there are still significant numbers of elderly demented patients who have many plaques in the neocortex, without neocortical tangles, or Lewy bodies. A few years after the original electron microscopy, plaques very similar to those of the human were found in aged primates of several species and in old dogs. These lesions differed from human only in that the dystrophic neurites did not contain paired helical filaments. Neuronal tangles are absent. The amyloid in the plaque cores reacted with antibodies prepared from human amyloid. The dog and monkey thus became the first available animal models of the disease, but they are not entirely satisfactory in that they are not convenient for laboratory work and that they both lack the neurofibrillary tangles. Aluminum came up in this regard when we found that rabbits treated intracisternally with aluminum salts developed fibrillar masses in certain neuronal groups. Ultrastructural studies, however, revealed that these fibers were identical to normal neurofilaments, and subsequent chemical analyses bore this out. PHF are quite different. Nevertheless, this finding stimulated research on the relationship of aluminum to Alzheimer’s disease, and this issue has not yet been settled.

Cell counting was greatly facilitated by the development of computerized image analysis. The image requires considerable manual editing because of the frequent crowding of cells, and the necessary elimination of vascular cells and artifacts. The technique demonstrated the expected major loss of large neurons in the Alzheimer neocortex amounting to about 35%. The difference between neuron counts in early onset Alzheimer’s disease and age-matched normals is highly significant and very obvious. In the more common late-onset cases, there is a smaller but still significant difference from age-matched normals. Not all morphometrists agree with this position. It was a few years later that the same instruments were used for analysis of neuron numbers as a function of normal aging. There it was found that the number of large neurons progressively decreased, but the number of small neurons increased about equally. It was therefore concluded that the large neurons were shrinking, rather than being lost as in AD. All these data are still debated, with at least one group even stating that there is no neuron loss in Alzheimer’s disease. Simple microscopic examination of cortex, even without counts, would seem to belie this last position.

In 1976 and early 1977, three neuropharmacologic papers were published almost simultaneously in Britain which together enormously increased investigative interest in AD. All demonstrated a highly significant diminution of choline acetyltransferase, thereby proving that the cholinergic system was particularly vulnerable in the disease. Soon thereafter, Whitehouse et al. showed that neurons of the basal nucleus of Meynert, the major source of cortical and hippocampal cholinergic projections, were markedly diminished in number and were frequently affected by tangles. It was not long before other neurotransmitter deficiencies were found, and corresponding neuronal losses demonstrated, as, for example, the locus ceruleus in regard to norepinephrine and the dorsal raphe associated with diminished serotonin. Transmitter deficiencies in the disease are probably caused by the loss of the neurons in which the particular molecule is synthesized.

Silver stains of the Alzheimer neocortex display apparently broken neuritic processes, and this has been recognized for several decades. It was in 1987 that Kowall and Kosik recognized that these fragments are abnormal dendritic processes. These are called neuropil threads, and they have been found to contain the same hyperphosphorylated tau which is the major component of paired helical filaments.

It has long been known that the hippocampal pyramids are very susceptible to tangle formation. The significance of these lesions in the entorhinal cortex was relatively recently emphasized, first by the findings of Hyman et al., and more recently by those of Braak. Hyman showed that the death of entorhinal layer 2 neurons by way of tangle formation leads to loss of the perforant pathway into the hippocampus. Disease in deeper layers of the entorhinal cortex diminishes the efferent target from the hippocampus, which itself is thus isolated. Braak established six stages in the progression of Alzheimer’s disease, based largely on the presence of tangles. Most observers agree that the earliest stage involves a few such lesions in the transentorhinal region extending in the second stage into entorhinal cortex and then increasing in hippocampal and neocortical areas through stage six. These stages can be correlated with clinical severity.

Synaptic terminals are only about a micron in diameter and therefore have been enumerated only since the advent of electron microscopy. More recently, immunocytochemistry and its quantitative evaluation by densitometry or by confocal microscopy have made this enumeration more convenient. These latter techniques depend largely on the presence of synaptophysin, which is an antigenic protein integral to the membrane of the synaptic vesicles. In fact, even electron microscopy has been largely dependent on the presence of these vesicles for recognition of the presynaptic bouton, although an older ultrastructural method utilized the phosphotungstic acid reaction on the postsynaptic thickening. All methods agree that there is a significant loss of synaptic terminals in the AD cortex. This change correlates more strongly with the ante-mortem psychometric data than does any other morphologic or biochemical finding. The loss of synapses is proportionally greater than the loss of neuronal cell bodies, and so one can infer that the synapses are lost before the cell body. Such a sequence might well be the result of deficient axoplasmic flow, as was first suggested on other grounds in 1967. There is evidence indicating that neuronal microtubules are destabilized because apolipoprotein E-4 instead of E-3 leaves the tau protein normally bound to microtubules accessible to abnormal phosphorylation. The inadequacy of the microtubule transport mechanism would lead to diminished axoplasmic flow whether the phosphorylated tau is formed into paired helical filaments or is simply present in solution rather than bound to tubules. Diminished axoplasmic flow would lead to dystrophic changes in neurites and to loss of the synaptic terminals. That destruction of synapses is tantamount to a loss of the normal connectivity upon which cognition is based.

The changes undergone by the brain in the process of normal aging explain, at least in major part, the differences between early-onset AD and the late-onset form. The older normal brain has lost synapses and has shrunken or lost pyramidal neurons. This still normal older brain is, therefore, closer to a threshold of minimal connectivity where signs of dementia would appear with only relatively little further loss due to disease.

Since the mid 1980s and the isolation and sequencing of β-amyloid by Glenner,6 the majority of laboratory research on Alzheimer’s disease has centered on this peptide and its precursor protein APP. This large protein has both trophic and toxic epitopes leaving the possibility that filamentous amyloid is but a bystander or marker. Another still unproven possibility is that intraneuronal amyloid or the amyloid gathered almost invisibly at the synapse leads to destruction of the latter and ultimately the cell body itself. Transgenic mouse models utilizing mutations of the amyloid gene with a variety of promotors are at this time becoming somewhat closer in their resemblance to the human disease.

Modeling experiments, be they spontaneous in the subhuman primate or transgenic manipulations of mice, are almost always attempts to mimic the morphologic changes of the human disease. Since it is now accepted that loss of synapses in the neuropil is the ultimate cause of dementia, a useful model must present this change. Morphology and morphometry remain important techniques in this regard as well as to further general understanding of the pathophysiology of Alzheimer’s disease.