Cortical Inflammation in Alzheimer Disease but Not Dementia With Lewy Bodies

Background  There have been no previous studies on the role of inflammation in the brain for the second most common dementing disorder, dementia with Lewy bodies.

Objective  To investigate the degree of cortical inflammation in dementia with Lewy bodies (DLB) compared with Alzheimer disease (AD) and control brains.

Design and Main Outcome Measures  Post-mortem tissue collection from a brain donor program using standardized diagnostic criteria. Brains collected from January 1, 1993, through December 31, 1996, were screened and selected only for the presence or absence of tau neuritic plaques. Results of immunohistochemistry for HLA-DR were quantified using area fraction counts. Counts were performed by investigators who were unaware of the diagnosis. Results were compared across groups using analysis of variance and posthoc testing.

Setting  A medical research institute in Sydney, Australia.

Patients  Eight brains with DLB and without the tau neuritic plaques typical of AD, 10 brains with AD and no Lewy bodies, and 11 nondemented controls without significant neuropathological features were selected from a consecutive sample.

Results  Compared with AD, DLB demonstrated significantly less inflammation in the form of HLA-DR–reactive microglia in all cortical regions (P<.001, posthoc). The level of inflammation in DLB was comparable to that seen in controls (P=.54, post hoc).

Conclusions  Inflammation appears related to the tau neuritic plaques of AD. Despite similar clinical presentations, therapeutic anti-inflammatory strategies are not likely to be effective for pure DLB.

ALZHEIMER DISEASE (AD) is the most common form of dementia and is characterized by loss of neuronal integrity with consequent memory impairment and cognitive decline. The major neuropathological hallmarks of the disease are senile plaques and neurofibrillary tangle formation, although inflammatory and immune mechanisms accompany the neurodegenerative process.¹ In fact, recent research² suggests that inflammatory mechanisms may cause a considerable amount of the tissue destruction that invariably leads to dementia. Cross-sectional and longitudinal epidemiological studies support this by demonstrating that nonsteroidal anti-inflammatory drugs prevent or delay the onset of dementia.^3,4 This work has given considerable hope to the idea of a readily available and inexpensive treatment for dementia and, more important, a potential preventive therapy.

There has been relatively little research regarding the specificity of this potential treatment for similar dementing disorders with variable underlying cellular abnormalities. Chronic inflammatory processes have been described previously in the brains of patients with other neurodegenerative conditions, including Parkinson disease, Pick disease, Huntington disease, parkinsonism dementia of Guam, amyotrophic lateral sclerosis, and Shy-Drager syndrome.^5-7 However, very little data exist regarding the role of inflammation in the second most common dementing disorder, dementia with Lewy bodies (DLB).⁸ Patients with DLB are difficult to distinguish clinically from patients with AD because of the similarity in their dementing illness.⁹

Investigation into the pathologic mechanisms leading to DLB is complicated by the presence of neuritic plaques characteristic of AD.¹⁰ In some cases, the amount of neuritic plaque is sufficient enough to warrant a concomitant diagnosis of AD. However, relatively few cases of DLB have significant tau neurofibrillary tangles,^11-14 suggesting different cellular mechanisms for both diseases. Our aim is to assess inflammatory microglia using an antibody against the major histocompatibility protein HLA-DR in a number of cortical regions from brains with DLB compared with brains with the tau neuritic plaques of AD and from nondemented control subjects without significant neuropathological features. This should establish whether inflammation is a common mechanism underlying the dementing process.

Permission for collection of brain tissue was given following the implementation of a regional brain donor program for neurodegenerative diseases in 1993. The program included prospective clinical evaluations matching the Consortium to Establish a Registry for Alzheimer's Disease (CERAD) clinical criteria¹⁵ and was approved by the ethics committees of the South Eastern and Central Sydney Areas Health Services, Sydney, New South Wales, and the University of New South Wales, Sydney. Brains were collected at routine or brain-only post-mortem examinations within 44 hours of death (Table 1). Twenty-nine brains were selected for study. Of these, 10 fulfilled clinicopathological criteria for definite AD¹⁶ and had stage 5 or 6 tau neuritic plaques,¹⁷ 8 fulfilled clinicopathological criteria for DLB⁸ and had no or minimal tau neuropathological features, and 11 were without neurologic or neuropathological disease. The brains were selected for uniform and nonoverlapping cellular abnormality, ie, Lewy bodies or neurofibrillary tangles. Six brains with DLB had plaques present within the cortex (Figure 1, B). Neocortical DLB was diagnosed in 7 of the 8 brains, whereas the remaining brain had limbic DLB.⁸ Case details, including diagnosis, dementia score, age, postmortem delay, and dementia duration, are presented in Table 1.

For neuropathological diagnosis, the brains were sectioned, and blocks were taken from limbic (hippocampus and anterior cingulate), frontal, parietal, temporal, and occipital neocortex and from basal ganglia, diencephalon, brainstem, and cerebellum. Brains with macroscopic infarction or head injury were excluded. Sections were cut and stained with the modified Bielschowsky silver stain to identify neuritic plaque and tangles, and immunohistochemistry examination was performed using a ubiquitin antibody (product No. Z0458, diluted 1:200; Dako, Glostrupp, Denmark) and an α-synuclein antibody (product No. 18-0215, diluted 1:200; Zymed Laboratories, San Francisco, Calif) to identify Lewy bodies and Lewy neurites, and tau II antibody (T5530, diluted 1:10,000; Sigma-Aldrich, St Louis, Mo) for the identification of neuritic plaque of AD. Standard peroxidase visualization was used with diaminobenzidine (Sigma-Aldrich) as the chromogen.¹⁸ All controls underwent clinical assessment for dementia and were without any significant neuropathological features.

Tissue blocks were taken from superior frontal, anterior cingulate, inferior temporal, hippocampus, and parahippocampal regions. Antigen retrieval using 4% aluminum chloride (AlCl₃)buffer was performed before immunohistochemical procedures using an antibody specific for HLA-DR (M775 mouse monoclonal anti–human HLA-DR antibody; Dako). This ensured that the maximum numbers of activated microglia were detected. Tissue blocks were rinsed in distilled water for several hours, followed by immersion in 4% AlCl₃ overnight. Blocks were microwaved (1000-W microwave) for 6 minutes on 80% power in 4% AlCl₃ and washed in Tris buffer, 0.1 mol/L (Tris and 0.1% azide, pH 7.4), for 15 minutes before being placed in 30% sucrose solution overnight. Sections 50 µm thick were cut and washed in 50% alcohol 3 times for 15 minutes per wash. This was followed by a further wash in 50% alcohol with 3% hydrogen peroxide(20 minutes) to quench endogenous peroxidase activity before being stained immunohistochemically as free-floating sections. Sections were incubated with 10% blocking serum (10% normal horse serum in Tris buffer) before addition of the primary antibody (diluted 1:2000 in Tris buffer) overnight at 4°C. Sections were removed from the refrigerator and washed in Tris–triton, 0.1 mol/L (3 times for 15 minutes per wash; 0.1% triton, no azide). Biotinylated secondary antibody (vector-biotinylated anti-mouse IgG) was applied and incubated at room temperature for 1 hour. After further washing 3 times for 15 minutes per wash, sections were incubated in avidin-biotin complex peroxidase(Vectastain Elite; Vector Laboratories Inc, Burlingame, Calif) for 1 hour. A final Tris–triton wash 3 times for 15 minutes per wash was performed in preparation for visualization using diaminobenzidine according to manufacturer's instructions. Sections were then mounted onto gelatinized slides and allowed to dry before being counterstained with cresyl violet, dehydrated through alcohol (70%, 90%, 2 times at 100% for 5 minutes per dehydration), cleared in xylene, and coverslipped with xylene-based mountant (DPX; BDH Laboratory Supplies, Poole, England). The specificity of the immunohistochemical reactions was tested by omitting the primary antiserum. No peroxidase reaction was observed in these test sections.

The HLA-DR immunoreactivity of reactive microglia was used as an index of inflammation and was measured by area fraction counts using an 11×11 eyepiece grid at ×200 magnification from the 2 areas containing the greatest density of staining in the gray and in the white matter.

Each slide for each brain was assessed, and the counts were expressed as a percentage of the total grid points. The values of 10 repeated counts on slides from the same brain at different times did not vary by more than 5%. Counts of the same slides from 4 brains by different investigators varied on average by 10%.

Statistical analysis was performed using commercially available software (Statview; Abacus, Berkeley, Calif). Two-way analysis of variance (ANOVA) was used to test for significant differences between diagnostic groups and cortical regions sampled. Fisher protected least square difference post hoc tests were used to identify the specific groups affected. P<.05 was accepted as the level of significance for all tests.

There was no significant difference between the mean age (P=.56) or postmortem delay (P=.51) among groups (ANOVA; Table 1). There was also no significant difference in disease duration (P=.44) or dementia severity (P=.41) (assessed using the Clinical Dementia Rating Scale) between the brains with AD and those with DLB (post hoc; Table 1).

The plaques were found in all brains with AD (by definition¹⁹) (Figure 1, A), in 6 of 8 brains with DLB (Figure 1, B), and in 3 of 11 elderly controls (data not shown). Only the brains with AD exhibited classic plaques with tau-positive neuritic infiltrates and an amyloid core (Figure 1, A-C). All brains with AD reached Braak and Braak stage 5 or 6, with the greatest numbers of plaques and tangles in the inferior temporal and hippocampal regions, as expected.¹⁷ No brain with AD had Lewy bodies within the cortex that could be detected using ubiquitin or α-synuclein immunohistochemistry.

The brains with DLB were selected for their lack of AD tau neuritic plaque; thus, tau immunoreactivity was rarely associated with plaque in these cases. However, 6 of the 8 brains with DLB exhibited plaque with ubiquitin- and α-synuclein–positive neurites (Figure 1, D). Classification of all brains with DLB were Braak and Braak stage 0 or 1.¹⁷ Therefore, none of the brains with DLB included in this study met neuritic criteria for AD.¹⁷ All brains with DLB had significant numbers of ubiquitin- and α-synuclein–positive Lewy bodies and Lewy neurites within cortical regions, with the greatest concentration of Lewy bodies in the anterior cingulate cortex (Figure 1, E) and the greatest concentration of Lewy neurites in the CA2-3 region of the hippocampus (Figure 1, F).

None of the controls had Lewy bodies or neurites within the cortex. Three controls had some plaques, although this was of the diffuse kind associated with normal aging and was seen in the eldest brains only (aged 85, 86, and 91 years).

Four cortical regions were analyzed, ie, superior frontal, inferior temporal, anterior cingulate, and entorhinal cortices, and the CA2-3 region of the hippocampus. In controls, more HLA-DR–positive microglia were found in the gray matter compared with the white matter (on average, 2 times more). There was no significant difference in the density of inflammatory microglia in any cortical gray matter region (P=.61, AN0VA; Figure 2, A). However, there was significantly less HLA-DR microglia in the white matter of frontal (P=.006), inferior temporal (P=.02), and anterior cingulate (P=.03) cortices than in the tracts between the hippocampus and parahippocampus (posthoc). Consequently, there were equivalent numbers of HLA-DR microglia in the white and gray matter of the hippocampus and surrounding cortex of controls.

In brains with AD, there was a distinctive distribution of HLA-DR immunoreactivity in gray matter, as described by others.^20-22 Reactive microglia were found in dense aggregates in the cortex (Figure 2, B). These aggregates appeared in a similar location and distribution to plaque (Figure 1, A). In brains with AD, a significantly increased number of HLA-DR immunoreactive microglia were observed throughout the cortex and hippocampus, in the gray and white matter, compared with DLB and control groups (P<.001, post hoc; Figure 3).

The distribution of microglia in brains with DLB was similar to that in controls (Figure 2, C), ie, a uniform distribution throughout the cortex and hippocampus without well-defined clusters. There was no significant increase in the density of reactive microglia in brains with DLB compared with controls (P=.54, post hoc; Figure 3). This contrasts with the significant increase found in brains with AD (Figure 2 and Figure 3).

The data described herein illustrate that only demented patients with significant tau neuritic plaques appear to have inflammation within the brain. In particular, patients with pure DLB exhibited no evidence of an inflammatory response underlying the cellular abnormalities causing their dementia. This contrasts significantly with the inflammatory response seen in patients with the tau neuritic plaques of AD, suggesting that significant cortical inflammation is more selective for this neurodegenerative condition. This has important implications for the use of anti-inflammatory drugs in dementia prevention, as approximately one third of patients fulfilling clinical criteria for AD have Lewy bodies at autopsy.⁸

In agreement with our study, a number of researchers have demonstrated a consistently higher expression of HLA-DR immunoreactive glia in patients with AD compared with age-matched controls.^1,5,22,23 The up-regulation is due to the clustering of microglia within tau-positive neuritic plaque.^1,24 This suggests that activation of inflammatory microglia is selective for the tau neuropathological feature in AD. However, a number of studies^5-7 have demonstrated a significant inflammatory response in neurodegenerative conditions that do not require tau neuritic plaques for diagnosis, including Parkinson disease, Pick disease, Huntington disease, amyotrophic lateral sclerosis, and Shy-Drager syndrome.

To attribute a role for brain inflammation in any neurologic disease, cases must be carefully selected using inclusion and exclusion criteria to account for overlapping cellular abnormalities. However, the selection methods of previous studies suggest that overlapping cellular abnormalities have not always been taken into account. For example, McGeer and coworkers⁷ reported a significant inflammatory response in demented patients with Parkinson disease; however, all of their patients had neuritic plaque and tangles consistent with a concomitant diagnosis of AD. Coexistent AD is not uncommon in patients with Parkinson disease²⁵ or Pick disease²⁶ and may have contributed to previous results demonstrating inflammation in these disorders.

Considerable debate remains whether DLB and AD can be differentiated clinically.⁹ Our results suggest that it will be important to do so as significantly different neuropathological features underlie these similar types of clinical dementia. Specifically, the brains with AD and DLB analyzed herein had a similar disease duration and dementia severity, but the brains with DLB had limited tau neuritic plaque and HLA-DR inflammatory glia, and the brains with AD had no Lewy bodies. Our previous volumetric analyses have shown that the hippocampus degenerates to a similar degree in DLB and AD^27,28; this may account for the similarities in clinical symptoms. However, clinical trials have shown differences in treatment efficacy between these disorders. Specifically, the use of tacrine hydrochloride for the relief of dementia appears particularly beneficial to patients with DLB but not those with AD.²⁹ As both disorders have early and severe degeneration of the cholinergic basal forebrain,³⁰ it has been proposed that the therapeutic difference is caused by a significantly greater cortical cholinergic deficit in DLB.³¹ Alternatively, such differences could also be due to the additional refractory inflammation associated with tau deposition in AD.

Our results suggest that brain inflammation may be specific to conditions with significant tau neuritic plaques. Neuritic plaques are common to AD and DLB; thus, inflammation and tau deposition appears to differentiate these degenerative dementias. Pathologic differences between these dementing disorders are supported by previous findings from our laboratory that patients with pure DLB have frontal atrophy²⁸ compared with the temporal cortical atrophy found in patients with AD.²⁷ That these similar dementia syndromes appear responsive to different drug therapies is further evidence that their pathogenesis significantly differs, warranting renewed vigor in devising strategies for their clinical discrimination. Differences between AD and DLB may be important clinically and pathologically in light of the proposed use of nonsteroidal anti-inflammatory drugs and other medications for dementia treatment and prevention.

Accepted for publication April 27, 1999.

This work was funded by the National Health and Medical Research Council of Australia, Canberra, and Parkinson's New South Wales, the Ian Potter Foundation, Melbourne, and the Clive and Vera Ramaciotti and Gerontology Foundations, New South Wales, Australia.

We would like to thank all the donors who made this work possible and James Pearse for technical support. We appreciate the assessment of these cases by various clinicians, including Tony Broe, FRACP, and Helen Creasey, FRACP, at the Centre for Research and Education on Ageing, Concord Hospital, and John Morris, FRACP, DM, and Mariese Hely, FRACP, at Westmead Hospital, and the help with pathological diagnoses by Jillian Kril, PhD, at the Centre for Research and Education on Ageing, Concord Hospital, New South Wales, Australia.

Reprints: Glenda M. Halliday, PhD, Prince of Wales Medical Research Institute, High Street, Randwick, New South Wales 2031, Australia (e-mail: g.halliday@unsw.edu.au).