The insular cortex plays important roles in a variety of regulatory mechanisms ranging from visceral control and sensation to covert judgments regarding inner well-being. The dementia of Alzheimer disease (AD) often includes behavioral dyscontrol and visceral dysfunction not observed in other diseases affecting cognition. This could be related to autonomic instability and to loss of the sense of self, and pathologic changes within the insula may play essential roles. The pattern of insular pathology of 17 patients with AD was examined and the severity of pathology was compared with that of the entorhinal cortex (EC), a region involved early in AD with reciprocal connections to the insula. Thioflavin S staining and Alz-50 immunostaining revealed that the insula carries a heavy burden of pathology in AD. Neurofibrillary tangles (NFTs) were largely confined to the deep layers of the cortex, whereas neuritic plaques (NPs) were distributed throughout the cellular layers and subcortical white matter. The density of NFTs, but not NPs, was highly correlated with the degree of EC pathology. However, NFTs were not seen in the insula until EC pathology reached a relatively advanced level. The density of insular NFTs varied according to architectonic type, with agranular cortex most affected, dysgranular cortex less affected, and granular cortex least affected. Thus, the insula is often involved in AD, and some of the behavioral abnormalities in AD may reflect insular pathology.
In its early stages, Alzheimer disease manifests itself as cognitive changes dominated by defects in memory and spatial orientation. These early cognitive deficits are likely the result of neuropathologic involvement of cerebral association cortices and limbic structures, especially the entorhinal cortex and other structures in and around the hippocampal formation. As the disease progresses, further behavioral and neurologic disintegration, including autonomic dysregulation, often emerges. This autonomic dysfunction worsens morbidity and probably foreshortens the lives of patients with this disease. The neurobiologic basis for the autonomic dysregulation in Alzheimer disease remains poorly understood. Recent efforts have highlighted some subcortical (1-4) and cortical (5, 6) correlates of the autonomic dysfunction. However, missing from the extant research is consideration of the insular cortex, which forms the topic of this report.
The insular cortex plays important roles in a variety of regulatory mechanisms ranging from visceral control and sensation to covert judgments regarding inner well-being (7-10). Focal pathology of the insular cortex by tumor, stroke, and epilepsy often produces gastrointestinal, vasomotor, cardiovascular, and other visceral disturbances (11-16). The leading causes of death in Alzheimer disease include myocardial infarction, cardiac failure, and bronchopneumonia (17-19). All of these failures of visceral organs may be triggered or exacerbated by autonomic dysfunction (20-23), and insular pathology may play a central role.
Multiple studies have suggested that the insula is pathologically involved in Alzheimer disease. Neuroimaging studies have revealed atrophy of the insula (24-26), whereas other studies have shown changes in insular blood flow (27), biogenic amine concentrations (28), muscarinic receptor densities (29), metabolic activity (30), monoamine oxidase-B activity (31, 32), and decreased RNA and protein content (33). Despite the strong clinical rationale and the many laboratory hints that the insula may be an important target of Alzheimer disease, neuropathologic changes within the insula in Alzheimer disease have not been described. We hypothesized that the insular cortex undergoes substantial pathologic change in Alzheimer disease.
The cortical sites at which neuropathology in Alzheimer disease initially appear are not randomly distributed. In particular, neurofibrillary tangles (NFTs), a hallmark pathologic change consisting principally of abnormally phosphorylated tau protein, are initially localized to the perirhinal region of the temporal lobe (34). As the disease progresses, tangles arise throughout much of the limbic cortex and can be observed in the isocortex only in the later stages of disease (35).
Regional differences exist not only in the timing at which NFTs first appear, but also in the density of these tangles throughout the course of the disease. Certain lobes such as the limbic lobe are much more heavily burdened with NFTs than are other lobes such as the occipital lobe. Furthermore, the density of NFTs is strongly linked to cortical cytoarchitecture. In particular, allocortex and periallocortex, including the entorhinal cortex and hippocampus, contain the greatest concentrations of tangles. Across the cerebral cortex, vulnerability to NFTs is correlated with cortical cytoarchitecture and with the connectional distance (i.e. the number of synapses removed) from the heavily affected limbic regions (36).
In macaques, the insula consists of 3 distinct but contiguous cytoarchitectonic regions-agranular cortex, dysgranular cortex, and granular cortex-arranged concentrically (37, 38). Furthermore, the cytoarchitectonic organizations of these 3 regions correlate with their connectional anatomy (39-41). In particular, the agranular cortex has strong reciprocal connections with the entorhinal cortex and is functionally part of the limbic lobe (42). In contrast, the granular cortex is connected predominantly with other neocortical areas and is functionally linked to somatomotor systems (42, 43). The dysgranular cortex, lying between the granular and agranular cortices, is an anatomic and functional intermediate between these 2 other cortical regions. Thus, if the human insula resembles the macaque insula, then it contains 3 regions that differ markedly in their cytoarchitecture and in their connectional distance to limbic cortices. Consequently, these 3 regions might differ in their vulnerability to pathology in Alzheimer disease. We hypothesized that the densities of NFTs and neuritic plaques (NPs) vary among the 3 cytoarchitectonic regions of the insula, with the agranular cortex most affected and the granular cortex least affected.
Materials and Methods
Human brain tissue from 5 control (ages ranging from 73-84 years, mean 77.2) and 17 Alzheimer disease cases (ages ranging from 62-89 years, mean 77.5) were used in this study (Table). For each of the brains used, the subjects or their families had given consent that the tissue would be used for research. All of the brains had autolysis times of less than 6 hours and were provided by the University of Iowa Deeded Body Program.
The diagnosis of Alzheimer disease was confirmed using the CERAD (Consortium to Establish a Registry for Alzheimer Disease) (44) and Khachaturian (45) criteria, which included a history of clinical dementia and the presence of NFTs and NPs in the hippocampal-parahippocampal and neocortical areas. The Alzheimer disease cases had durations of clinical dementia ranging from 2 to 10 years.
As shown in Figure 1, the insula lies at the base of the Sylvian fissure and consists of a series of gyri arranged in the shape of a fan with its apex pointed anteriorly and ventrally. The insula's boundaries are demarcated by the circular sulcus. Underlying the insular gyri are the claustrum and putamen (40, 46).
The left or right insula was exposed by removing the opercular portions of the frontal, parietal, and temporal lobes (47, 48). The insula was then resected en bloc by placing a circumferential incision external to the circular sulcus (Fig. 1). The insular cortex, subcortical white matter, and a portion of the underlying basal ganglia were removed as a single block. From the same brains, a block of parahippocampal gyrus (49), containing the entorhinal cortex, was also resected. The blocks were immersed for 1 week in fixative containing 3% paraformaldehyde and 15% picric acid. The blocks were then cryoprotected by immersion in 30% sucrose, frozen in dry ice, and stored at -70°C until further processing.
Each insula and parahippocampal gyrus was cut into multiple series of sections with a sliding microtome in the coronal plane into 50-μm-thick sections. An insula sectioned in this way typically yielded 500 to 600 sections. A subset of the sections (2 in every 10) was mounted onto slides and stained with thionin for Nissl substance or with thioflavin S for NFTs and NPs.
Other series of insular sections (1 in every 20) were immunohistochemically stained for Alz-50, an antigen expressed in brains affected with Alzheimer disease and accompanying the formation of NFTs (50). The immunohistochemical methods have been described previously (51). Briefly, the sections were first incubated in the primary antibody, mouse anti-A68 (Alz-50, monoclonal, 1:1000; Abbott Pharmaceutical, Chicago, IL), and dissolved in normal serum containing 0.4% Triton X-100. After washing, the sections were incubated in the secondary antibody (biotinylated goat anti-mouse, 1:500) containing 0.1% Triton X-100. Antibody visualization was achieved using biotin-avidin complex (Jackson ImmunoResearch, West Grove, PA). For specificity control, one of the incubation steps was omitted, resulting in a complete suppression of the immunostaining.
Quantification of Alzheimer Disease-Related Pathology
To quantify Alzheimer disease-related pathology, a subset of the insula sections stained with thioflavin-S was viewed through a 20× microscope objective for quantification of NFTs or through a 4× microscope objective for quantification of NPs. Every fifth thioflavin-S-stained section was used so that the data were typically derived from approximately 12 sections along the anterior-posterior axis of the insula. The image from the Nikon Optiphot-2 microscope (Nikon Inc., Melville, NY) was projected onto a Hitachi SuperScan Elite 17 color monitor (Hitachi, Ltd., Tokyo, Japan) and transferred to a Micron Millenia LXR computer using a Diagnostic Instruments color video camera (Diagnostic Instruments, Inc., Sterling Heights, MI). NFTs within the projected area of 0.79 mm2 and NPs within the projected area of 4.86 mm2 were counted using the point counting program of StereoInvestigator software (MicroBrightField, Inc., Colchester, VT). NFTs and NPs were counted through the depth of the 50-μm-thick sections, which, after processing, had shrunk to approximately 12 μm in thickness. For determination of NFT density, the sampled region contained only insular gray matter, whereas for NP density, the sampled region contained both the gray and white matter of the insula. Each insula section was randomly sampled in its superior third, middle third, and inferior third, and the 3 scores were averaged for each section. Thus, NFT density and NP density were determined at approximately 36 sites per insula (3 sites on each of 12 sections along the rostrocaudal length). The locations for sampling were obtained by electronically outlining the insula's boundaries with the StereoInvestigator software and then using the software to randomly choose sampling sites within the outlined area.
Comparison of Insular and Entorhinal Pathology
One goal of this study was to compare the degree of Alzheimer disease-related pathology in the insula with that of the entorhinal cortex, a brain region affected early and severely by Alzheimer disease (34, 36, 49). To accomplish this, sections of parahippocampal gyrus stained with thioflavin-S were examined and the grade of entorhinal pathology for each case was determined using a variation of the staging scheme proposed by Braak and Braak (34). The grade of entorhinal pathology was assigned according to the distribution of NFTs as follows: 0 = no or only rare NFTs (control); 1 = NFTs confined to layer II; 2 = NFTs confined to layers II and IV; 3 = NFTs confined to layers II, III, and IV; and 4 = NFTs in all cell layers. Severity of insular pathology was compared with the level of entorhinal pathology, as explained subsequently.
Comparison of Alzheimer Disease Pathology Among the Insula's Cytoarchitectonic Fields
A principal hypothesis of this study is that the density of Alzheimer disease-related pathology within the insula depends on cytoarchitecture. The arrangement of architectonic regions of the insula has been well described in nonhuman primates (37-40). However, the arrangement has been described in only a single human case (37) and has never been mapped. Thus, using the 4 Nissl-stained control insulas and an additional control brain stained by the Gallyas cell stain (52), we charted the pattern of agranular, dysgranular, and granular cortices within the insula. We then produced a composite map depicting the typical arrangement of the 3 architectonic types within the human insula.
With this map as a guide, we identified the boundaries among the agranular, dysgranular, and granular insular cortices in the thionin-stained sections of the Alzheimer disease cases. In many cases, identification of the boundaries between architectonic regions of the insula in the Alzheimer cases was greatly facilitated by the control map, because the loss of neurons, driven by Alzheimer disease, often partially obscured the cortical cytoarchitecture. The relative sparing of layer IV in Alzheimer disease was also fortuitous to make such distinctions, because this layer's appearance is a major determinant in partitioning the insular cortex into granular, dysgranular, and agranular sectors. With the boundaries among cytoarchitectonic regions delineated in the thionin-stained sections, we likewise identified the boundaries in the adjacent sections stained with thioflavin-S. Densities of NFTs and NPs were measured at 3 randomly chosen sites within the boundaries of each cortical type for each section examined, as described previously.
For both the Alzheimer disease cases and the control cases, the density of NFTs from each of the 3 sites sampled on each section from each case was averaged to yield a mean NFT density for each case. The same procedure was followed for density of NPs. For the Alzheimer disease cases, least-squares linear regression analyses were conducted to determine whether the density of NFTs and NPs were correlated with the number of years of clinical dementia. The F test was used to assess goodness-of-fit of the regression lines. Correlation coefficients (r) were calculated by simple linear regression. To determine whether the densities of NFTs and NPs were correlated with the level of entorhinal pathology, nonparametric rank correlations were conducted, and the Spearman rank correlation coefficients (rs) were calculated (53). Significance of the Spearman rank correlations was tested using at distribution.
For analysis of differential pathology within agranular, dysgranular, and granular regions of the insula, densities of NFTs and NPs obtained from 3 sites within each region of each examined section were averaged to yield a mean density of NFTs and NPs by region for each case. These densities of NFTs and NPs by region were analyzed by repeated-measures analysis of variance (ANOVA) with architectonic region as the grouping factor and densities of NFTs and NPs as the repeated measures (54, 55). Post hoc comparisons among the 3 regions were conducted using the Newman-Keuls procedure.
The Insular Cortex Is Pathologically Involved in Alzheimer Disease
In all 17 cases of Alzheimer disease, the insular cortex had clear evidence of Alzheimer disease-related pathology. Sixteen of the 17 cases had NPs, whereas 14 of the 17 cases had NFTs. As shown in Figure 2, in many cases, the insula carried a heavy burden of both NFTs and NPs. Although NFTs were present in all depths of the insular cortex, they were often most heavily concentrated in the deeper cortical layers. In contrast, NPs were distributed uniformly throughout the cell layers and subcortical white matter. Furthermore, Alz-50 immunostaining labeled neuronal cell bodies and neuritic processes in all cell layers of the insula in 12 of the 17 cases of Alzheimer disease. In contrast to the cases with Alzheimer disease, the control cases had no NPs and only very rare and isolated NFTs. Thus, the insula is severely pathologically affected in many cases of Alzheimer disease.
Density of Insular Neurofibrillary Tangles, but not Neuritic Plaques, Is Correlated With Years of Clinical Dementia
Figure 3 shows the relationship between NFT density and years of clinical dementia, as well as the relationship between NP density and years of clinical dementia. The density of insular NFTs was positively correlated with the number of years of clinical dementia (r = 0.71; p < 0.005). The greater the number of years demented, the greater the density of insular NFTs. In contrast, the density of insular NPs was not correlated with the years of clinical dementia. Neither the density of NFTs nor of NPs was correlated with age.
Alzheimer Disease-Related Pathology in the Insula Increases as Pathology in the Entorhinal Cortex Increases
The entorhinal cortex is an early and highly vulnerable target of Alzheimer disease pathology. Therefore, comparison of pathologic burden in the insula with that of the entorhinal cortex provides insight into the time-of-onset and progression of insular pathology in the course of the disease. As shown in Figure 4, both the nature and the burden of Alzheimer disease pathology within the insula were closely related to the grade of entorhinal pathology. In particular, the density of NFTs within the insula increased as the grade of entorhinal pathology worsened. However, NFTs did not occur in the insula until pathologic changes in the entorhinal cortex were relatively advanced, with NFTs present in layers II and IV of the entorhinal cortex (grade 2 of entorhinal pathology). Likewise, Alz-50 immunohistochemical staining labeled cell bodies and neuritic processes of the insula only sparsely at grade 1 entorhinal pathology but progressively more densely as the grade of entorhinal pathology increased. In contrast, NPs were present within the insula at all grades of entorhinal pathology, and the density of NPs did not increase as the grade of entorhinal pathology worsened.
These relationships are shown quantitatively in Figure 5. The density of insular NFTs was significantly correlated with the grade of entorhinal pathology (rs = 0.74, p < 0.0005). In contrast, the density of insular NPs was not significantly correlated with the grade of entorhinal pathology (rs = 0.046, p > 0.30).
The Human Insula Contains 3 Distinct Architectonic Types
Five control human brains stained with thionin or the Gallyas cell stain were examined to determine the presence and distribution of different cytoarchitectonic regions within the insula. As shown in Figure 6, we found that the human insula, like the monkey insula (37, 40, 48), is composed of 3 distinct cytoarchitectonic regions. Furthermore, like the monkey insula, these 3 regions are arranged concentrically with increasingly differentiated cortex in the radial direction.
The rostroventral portion of the human insula consists of agranular cortex. This periallocortex contains no aggregates of granule cells and lacks cortical layers II and IV (Fig. 7). It consists of only 3 cell layers, the innermost layer of which is contiguous with layer V of the dysgranular cortex dorsally and with the underlying claustrum.
Radial to and continuous with the agranular cortex is the dysgranular cortex. In this proisocortex, granule cells begin to cluster but form only incipient cortical layers II and IV. Neurons of layer VI contiguous with the underlying claustrum are far less common in the dysgranular cortex than in the granular cortex. Figure 6 shows that dysgranular cortex comprises the majority of the human insula.
The most dorsal and posterior portion of the insula is composed of granular cortex. This isocortex demonstrates the highest degree of cortical differentiation within the insula. In the granular cortex, layers II and IV are conspicuous, fully demarcated, and continuously defined. Throughout the extent of the granular layer, there is a clear demarcation between layer VI and the underlying white matter with no interdigitation of cells between granular cortex and claustrum.
Severity of Alzheimer Disease Pathology Varies Among the Insula's Cytoarchitectonic Regions
As shown in Figure 8, the density of NFTs within the insula depended strongly on cytoarchitectonic type (F2,48 = 3.96, p < 0.05). The density of NFTs was significantly greater in the agranular cortex than in either dysgranular (p < 0.05) or granular cortex (p < 0.05). Furthermore, the density of NFTs was significantly greater in dysgranular cortex than in granular cortex (p < 0.05). Thus, NFTs became increasingly dense in increasingly primitive forms of insular cortex. As was true for NFTs, NPs were most dense in agranular cortex and least dense in granular cortex. However, regional differences were not as strong for NPs as they were for NFTs, and regional differences in NP density were not statistically significant. Figure 9 provides examples of insular agranular cortex, dysgranular cortex, and granular cortex, all derived from a single individual with Alzheimer disease. Note that the degree of Alzheimer disease-related pathology, especially the density of NFTs, varies markedly among the cytoarchitectonic regions, with agranular cortex most affected.
This study demonstrated that the insular cortex is a prominent target of Alzheimer disease. In many cases of Alzheimer disease, the insula is heavily burdened with NFTs, NPs, and Alz-50-positive neuronal cell bodies and neurites.
These results predict that the neurologic and behavioral functions subserved by the insula are impaired in Alzheimer disease and that insular dysfunction is part of the disease's symptomatology. This is indeed the case. Clinical and experimental studies have demonstrated that the insula plays a role in memory, drive, affect, gustation, olfaction, and autonomic control. Each of these spheres of neurologic function can be impaired in patients with Alzheimer disease.
The insula has strong reciprocal connections with primary olfactory cortex (38-40, 56). This is particularly true of the agranular insula (42), the most vulnerable region of the insula to Alzheimer disease. Olfaction is substantially impaired in many cases of Alzheimer disease (57-60), and dysosmia may be one of the earliest signs of the disease. Some have suggested that olfactory dysfunction may be an early marker of Alzheimer disease and that olfactory tests may be clinically useful to discriminate Alzheimer disease from other brain disorders (57, 61-64). Olfactory deficits in patients with Alzheimer disease are the result, not principally of rhinologic pathology, but of brain pathology (65). Impairment of odor detection in Alzheimer disease is likely largely the result of pathology of the olfactory bulb, whereas impairment of odor discrimination and identification most likely reflect dysfunction at the cortical level (66). The insula might be the key cortical sight from which olfactory dysfunction arises in Alzheimer disease.
The insula also plays an important role in the sense of taste (39, 67). As part of the primary cortical gustatory area, the insula contains neurons that respond specifically to gustatory stimuli (68, 69). For this reason, pathology of the insula can produce gustatory dysfunction. For example, focal epilepsies arising from insular cortex may include gustatory auras as part of their semiology (15, 70), and electrical stimulation of the insula may produce gustatory sensations (70). Although gustation has not been extensively investigated in Alzheimer disease, several studies have demonstrated that the sense of taste is impaired in this disease (71, 72). The gustatory impairments can occur early in the course of Alzheimer disease and represent an associative agnosia in which gustatory thresholds are preserved whereas gustatory identifications are impaired (71). This suggests that the abnormalities in gustation in Alzheimer disease are mainly the result of dysfunction of the central rather than peripheral nervous system. The insula is the most likely focus of pathology underlying the gustatory dysfunction of Alzheimer disease.
Although olfaction and gustation are mediated by the insula and abnormal in Alzheimer disease, impairment of other functions subserved by the insula probably have an even greater clinical impact in Alzheimer disease. In particular, the insula integrates visceral sensory and motor functions and plays a major role in regulating autonomic responses (7, 12, 39, 48). Insular pathology can disturb cardiac function, blood pressure, respiratory patterns, and intestinal motility (9, 12, 14, 73). Alzheimer disease is malignant; patients with the disease have a higher mortality rate than age-matched controls (17, 74). Although part of this increased mortality is undoubtedly the result of dementia, another portion is likely the result of visceral dysfunction and loss of homeostasis (75). Insular pathology probably contributes strongly to the increased rates of cardiac failure, myocardial infarction, bronchopneumonia, and other life-threatening disturbances of visceral function accompanying Alzheimer disease (18, 19, 76, 77).
The insula joins a growing list of brain structures in which Alzheimer disease pathology can lead to autonomic dysfunction. Brainstem autonomic nuclei are heavily and selectively affected by Alzheimer disease (3). In particular, the dorsal motor nucleus of vagus, nucleus tractus solitarius, and intermediate reticular zone are targets of Alzheimer disease and play a critical role in autonomic control of visceral functions (3, 4). Another subcortical site of Alzheimer disease pathology is the periaqueductal gray matter, which initiates and coordinates visceral responses to external and internal stimuli and which is often heavily riddled with NPs in Alzheimer disease (2). Likewise, the parabrachial nucleus plays an important role in the integration of visceral and nociceptive information and carries a heavy burden of NFTs in Alzheimer disease (1). In addition to these subcortical structures, several cortical regions, besides the insula, play important roles in autonomic control and have extensive pathology in Alzheimer disease. In particular, the orbitofrontal cortex (6) and the ventromedial frontal cortex (5) have strong hypothalamic connections that are disrupted by dense NFTs within the projecting neurons of these cortices in Alzheimer disease. Thus, it is likely that insular pathology converges with pathology of multiple cortical and subcortical structures to disrupt and disintegrate autonomic control and visceral sensory and motor functions.
Because the insula receives data from all sensory modalities (38) and has strong connections with both the limbic and autonomic systems (8, 39, 40, 78), the insula creates an awareness of the physical self across time. Thus, the insula integrates external and internal sensations with emotion and memory to create a perception of the self and of how the self feels (7, 79). The dementia of Alzheimer disease often includes behavioral dyscontrol not observed in other diseases that affect cognition. This could be the result of a loss of the “sense of self,” and insular pathology may play an essential role.
Probably because of its unique linkage to the combination of olfaction, gustation, visceral sensation, self-awareness, and emotion, the insula plays an essential role in recognizing and experiencing disgust (80-82). The anteroventral portion of the insula is particularly implicated in the emotion of disgust (11, 83). As we have shown, this same portion of the insula contains agranular and dysgranular cortex and is the most pathologically involved portion of the insula in Alzheimer disease. The emotion of disgust has not been studied in Alzheimer disease. However, anyone with experience caring for patients with Alzheimer disease knows that these patients often engage in behaviors such as producing and eating unpalatable food concoctions that most healthy people would find disgusting. Impairment in the ability to experience and recognize disgust may be another important deficit of Alzheimer disease with a neural basis in the insula. Functional magnetic resonance imaging, tests of ability to recognize and experience disgust, and quantitative neuropathology could be used together in future studies to examine the impact of Alzheimer disease on the recognition and experience of disgust and the relative role of insular pathology in the impairment of this emotion (80-83).
This study demonstrated that the insula is riddled with NFTs and NPs in many cases of Alzheimer disease. Furthermore, the density of insular NFTs is closely tied to the grade of entorhinal pathology. Insular pathology may be correlated with entorhinal pathology because entorhinal pathology may “pace” pathology elsewhere, including the insula. Consistent with this notion are the facts that the insula has reciprocal connections with the entorhinal cortex and that NFTs were never observed within the insula until the grade of entorhinal pathology reached or exceeded grade II. This is precisely the scenario that might emerge if the entorhinal pathology signaled or triggered insular pathology through a neuron-to-neuron “spread” of disease. Alternatively, the entorhinal cortex and insula might be independently responding to a common condition or trigger that induces pathology, and NFTs may emerge earlier in the entorhinal cortex than in the insula only because the entorhinal cortex is more vulnerable than the insula. Our finding that the density of neuritic plaques is independent of the grade of entorhinal pathology strengthens this notion.
Cytoarchitectonic regions within the insula varied markedly in their burdens of NFTs. In particular, agranular cortex was more severely affected than dysgranular cortex, which was more severely affected than granular cortex. Although this is the first study to demonstrate regional differences in NFT density within the insula, multiple studies have demonstrated such differences among other cortical regions (36, 84-86). Furthermore, most previous studies have reported that limbic cortices, especially the entorhinal cortex and the allocortical and periallocortical regions associated with it, are the most severely affected regions in Alzheimer disease, whereas neocortical sensory and primary association regions are far less vulnerable (5, 36, 87, 88). Our findings in the insular cortex corroborate this principle. The agranular cortex of the insula has strong reciprocal connections with the entorhinal cortex, is part of the paralimbic belt (38-40, 42), and carries the heaviest burden of NFTs within the insula. In contrast, the granular cortex has strong connections with primary somatosensory areas, is isocortical in structure, and carries the lightest burden of NFTs within the insula. The dysgranular cortex is intermediate in its structure, location, and connections and correspondingly carries an intermediate burden of NFTs. Thus, our findings of regional differences in pathology in the insula are strongly in keeping with the observations of others that the cortical burden of NFTs is a function of cortical structure and connectional anatomy (34-36, 84-86). The biologic basis for these marked regional differences in vulnerability remains a mystery and may provide the key to understanding the pathogenesis of Alzheimer disease.
Descriptions of Alzheimer disease symptomatology often stress memory and cognitive deficits because they play such an important role in the interactions of humans with each other and their environment. These deficits can be profoundly disabling but constitute only part of the problem in Alzheimer disease. Patients with Alzheimer disease experience a host of neurologic abnormalities, and involvement of the insula no doubt adds to this burden. In Alzheimer disease, insular pathology, along with changes in brainstem nuclei (3, 6), orbitofrontal (6), and medial frontal (5) cortices, probably corrupt concepts of self and well-being, alter visceral sensations, dysregulate autonomic functions, and interfere with decision-making and social interactions. Even if memory remained intact, Alzheimer disease would constitute an illness that substantially impairs quality of life.
The authors thank Mr. Paul Reimann for photographic assistance.