Deposition of β-Amyloid Subtypes 40 and 42 Differentiates Dementia With Lewy Bodies From Alzheimer Disease

Background  Alterations in the metabolism of the amyloid precursor protein and the formation of β-amyloid (Aβ) plaques are associated with neuronal death in Alzheimer disease (AD). The plaque subtype Aβ_x-42 occurs as an early event, with Aβ_x-40 plaques forming at a later stage. In dementia with Lewy bodies (DLB), an increase in the amount of cortical Aβ occurs without severe cortical neuronal losses.

Objective  To advance our understanding of the natural history of Aβ in neurodegenerative diseases.

Design  We evaluated the expression of Aβ_x-40 and Aβ_x-42 in DLB using monoclonal antibodies and immunohistochemical techniques in 5 brain regions. The data were compared with those elicited with normal aging and from patients with AD.

Setting and Patients  A postmortem study involving 19 patients with DLB without concurrent neuritic degeneration, 10 patients with AD, and 17 aged persons without dementia for control subjects.

Results  The Aβ plaques were more numerous in patients with DLB than in controls in most brain regions, although the Aβ_x-42 plaque subtype was predominant in both conditions. Overall, Aβ_x-42 plaque density was similar in patients with DLB and those with AD, but Aβ_x-40 plaques were more numerous in persons with AD than in those with DLB. The ratio of Aβ_x-40 to Aβ_x-42 plaques was significantly reduced in persons with DLB compared with patients with AD.

Conclusions  The Aβ plaques were more numerous in patients with DLB than persons with normal aging, but the plaque subtypes were similar. The relative proportion of the 2 Aβ plaque subtypes in DLB is distinguishable from that in AD.

THE MOLECULAR basis of neuronal degeneration in Alzheimer disease (AD) is not fully understood, but recent attention^1-14 has focused on the importance of the role of different-length β-amyloid (Aβ) peptides. Because there is an increased amount of the subtype Aβ_x-42, but not of Aβ_x-40,⁹ in presymptomatic patients with genetic forms of AD, including Down syndrome and presenilin 1 AD, Aβ_x-42 deposition is considered¹² an early event in the AD process, with Aβ_x-40 plaques occurring later. Both Aβ subtypes are present in the brains of older patients with Down syndrome and in patients with AD.^9,13-15

The biological factors associated with the accumulation of the different Aβ subtypes in degenerative diseases are not well understood. Brain specimens from patients with familial AD, however, show greater amounts of Aβ_x-42 than do those from patients with sporadic AD.^13-15 Compared with that in patients with sporadic AD, the ratio of Aβ_x-42 to Aβ_x-40 is also increased in patients with AD with mutations of the amyloid precursor protein (APP) on chromosome 21 and presenilin 1 mutations on chromosome 14.^13-15

Recent evidence^16,17 suggests that intraneuronal organelles are important in the Aβ cascade and in the formation of the Aβ subtypes. The subtypes Aβ_x-42 and Aβ_x-40 are formed in the endoplasmic reticulum and the Golgi apparatus, respectively.^16,17 Despite these major breakthroughs, the biological trigger for the Aβ cascade is unknown. Also unknown is whether the process of Aβ formation from APP involves an identical biochemical sequence of events in all Aβ-forming diseases.

The study of Aβ in other degenerative diseases may advance our understanding of the role of Aβ in AD and in neuronal degeneration. Dementia with Lewy bodies (DLB) is an increasingly recognized^18-21 subtype of dementia. Although DLB frequently occurs concurrently with AD changes, many patients have pure DLB.²² In patients with DLB not meeting pathological criteria for AD, Aβ plaque (AβP) deposition in the cortex is often increased compared with that in age-matched controls.²³ In patients with DLB, however, cortical neuronal loss is minimal.^24-26 Differences in Aβ_x-40 and Aβ_x-42 formation between AD and DLB suggest that the biochemical processes that lead to Aβ formation may differ. To address this issue, we examine the relationship between DLB, AD, normal aging, and the different lengths of Aβ.

We examined the brains of 19 patients (mean age, 75.0 years at death) (Table 1) who met consensus pathological criteria²⁷ for transitional or neocortical DLB. Specimens from 12 patients were obtained from the Neurology Brain Bank, MCP-Hahnemann University, Philadelphia, Pa, and specimens from 7 patients were obtained from the Department of Pathological Sciences University of Manchester, Manchester, England. Neurons were identified as containing Lewy bodies if inclusions had the morphologic appearance of a Lewy body and a nucleus was present. Cases were identified using hematoxylin-eosin stains and confirmed using ubiquitin and α-synuclein immunohistochemical stains. This group of patients had a varying number of diffuse AβPs shown by Aβ immunohistochemical techniques and silver stains (described later). The brain specimens lacked other notable neuropathological features; in particular, they did not contain cortical neurofibrillary tangles. All had infrequent plaques by criteria of the consortium on DLB international workshop,²⁸ and Braak stages²⁹ ranged from I to IV. Most would be categorized as having a low likelihood that their dementia is caused by AD.³⁰ No patients with DLB met criteria of the National Institute on Aging and the Reagan Institute Working Group on Diagnostic Criteria for the Neuropathological Assessment of Alzheimer's Disease³⁰ for a high likelihood that dementia was related to AD. Although parkinsonism developed in some patients, all patients with DLB had dementia as the first symptom. No other notable neurologic diagnoses were identified. The mean duration of disease was 7.6 years, and all patients but 1 MCP-Hahnemann University case had advanced dementia, were in nursing homes, and required assistance for all activities before death. The other patient had moderate disease severity at death. The disease severity was not known for the 7 patients from the University of Manchester.

We examined brain specimens from 10 patients from MCP-Hahnemann University, Philadelphia, who met clinical criteria of the National Institute of Neurological Disorders and Stroke and the Alzheimer's Disease and Related Disorders Association³¹ for possible or probable AD. All had frequent neuritic plaque scores according to criteria of the Consortium to Establish a Registry for Alzheimer's Disease²⁸ and Braak²⁹ stages V or VI. All met criteria of the National Institute on Aging–Reagan Institute Working Group³⁰ for a high likelihood that their dementia was related to AD. All patients lacked intracortical or nigral Lewy bodies. The mean age was 74.5 years at death. The mean disease duration was slightly longer in patients with AD (9.5 years) than in those with DLB, but these differences were not significant (P=.29). All except 1 patient died with advanced disease. This patient had moderately severe manifestations of the disease at death. None had significant concurrent neurologic diagnoses.

We examined brain tissue from 17 control subjects who died without significant dementia or neuropathologic diagnoses. Cognitive status was confirmed by medical records review. The mean age at death was 75.9 years. Braak²⁹ stages were I or II, and staging according to the consortium on DLB international workshop²⁸ showed that neuritic plaques were infrequent. None had cortical or nigral Lewy bodies.

Genotyping for APOE was performed on brain tissue using a polymerase chain reaction protocol.³²

We obtained coronal sections of the middle frontal gyrus, medial temporal lobe (CA1 and CA3 sectors of the hippocampus and parahippocampal gyrus [PHG]) at the level of the lateral geniculate nucleus, and cerebellar hemisphere lateral to the dentate gyrus. Tissue blocks were embedded in paraffin and cut to a thickness of 6 µm.

Sections were stained with antibodies directed against Aβ_x-40 (6A; monoclonal, 1:1000 dilution) and Aβ_x-42 (11C; monoclonal, 1:1000 dilution). Both antibodies—produced using standard methods—are C-terminal specific. Briefly, BALB/c mice (Japan SLC, Shizuoka, Japan) were immunized with the synthetic peptides with the sequence of either CVGGVV or CGVVIT, which corresponded to Aβ_36-40 or Aβ_38-42. The aminoterminal cysteine was added to the synthetic peptides for conjugation with keyhole limpet hemocyanin (Wako Chemicals, Osaka, Japan). After several immunizations, spleen lymphocytes from the immunized mice were fused with myeloma cells to generate the hybridoma. Following 2 limited dilution series, we screened the 2 monoclonal antibodies 6A and 11C for specificity by enzyme-linked immunosorbent assays using relevant synthetic peptides¹⁴ and Western blot analysis (Figure 1). Western blot specimens included β001, a rabbit polyclonal antibody against the Aβ_1-42 peptide, and a monoclonal antibody (2Fi) directed against Aβ_18-28, in addition to the 6A and 11C antibodies. These antibodies recognize both Aβ_x-40 and Aβ_x-42 species. A total of 100 ng of peptides was used for each lane. Sodium dodecyl sulfate polyacrylamide gel electrophoresis was carried out under reducing conditions on all specimens, and proteins on gels were transferred on PVDF membrane (Hybond-P, Amersham Life Sciences, United Kingdom). The membrane was blocked with 3% gelatin and incubated with the monoclonal antibody, followed by the horseradish peroxidase–conjugated second antibody. The color was developed with 4-chloro-1-naphthol. The protein concentration was quantitated by measuring the absorption at 562 nm using an assay kit (BCA, Pierce, Rockford, Ill).

In addition, each antibody was examined for the appropriate pattern of immunoreactivity using immunohistochemical technique on paraffin sections of brain from patients with AD. We used the biotin-streptavidin technique with diaminobenzidine and a light hematoxylin counterstain. All sections were pretreated with formic acid. Tissue designated "negative control" included adjacent sections where nonimmune serum replaced the primary antibodies. Tissue designated "positive control" included the cerebral cortex of patients with advanced AD known to have an abundance of both Aβ_x-40 and Aβ_x-42.

All Aβ data were obtained by 2 observers (C.F.L. and Brendan O'Connell) who were blinded to the subjects' diagnosis during data acquisition. We obtained data from the middle frontal gyrus, the hippocampal CA1 and CA3 regions, the PHG, and the cerebellar hemisphere. To make our data comparable to those of other Aβ studies, we measured AβP densities and calculated the percentage of area with Aβ (amyloid burden) from the field with the maximal AβP deposition within each region of interest.^8,14 For density data, all AβPs at least as large as a small neuron were counted at a magnification of ×10. Each field was 3.5 mm². Densities in each region of interest were expressed as the number per square millimeter.

To determine the amyloid burden, we used a previously described sampling procedure³³ but captured images using a digital camera (SenSys; Photometrics Ltd, Munich, Germany) attached to a light microscope (Precision Instrument Division, Olympus Corporation, Lake Success, NY). The percentage of each field stained by Aβ was calculated using a semiautomated computer program (Image-Pro; Media Cybernetics, Silver Spring, Md) and sampling 0.6-mm² fields.

For each specimen, the Aβ_x-40:Aβ_x-42 ratio was calculated in each region using both Aβ sampling methods.

Data were analyzed using an analysis of variance. Data differences that were significant at the P≤.05 level were further analyzed using the Tukey-Kramer method to determine which groups differed from each other. This method adjusts the P value to allow for multiple comparisons.

Microscopic examination in affected regions showed that the Aβ antibodies detected AβPs in regions consistent with their known distribution in AD. The subtype Aβ_x-42 was common in all patients with AD and all patients with DLB except 1 but was more variable in controls. The same brain regions were preferentially affected in all groups, with the greatest number of AβPs in the frontal cortex, the fewest number in the PHG (CA sectors), and the lowest number in the cerebellum. In patients with DLB, Aβ_x-42 plaques sometimes appeared smaller and more irregular than those observed in patients with AD. As expected, our Aβ_x-40 antibody stained fewer AβPs than the Aβ_x-42 antibody. Overall, Aβ_x-40 was rare in control specimens and those from patients with DLB and variable in patients with AD. In addition to AβP, patients with AD showed Aβ_x-40 immunostaining of meningeal and intraparenchymal blood vessels.

When Aβ_x-40 densities were compared (Table 2), overall differences were significant between control specimens (Figure 2, A), specimens from patients with DLB (Figure 2, C) and those from patients with AD (Figure 2, E) in all regions except the cerebellum. The F scores revealed that the group with AD had greater Aβ_x-40 plaque densities than either the DLB group or the control group. In the cerebellum, statistical analysis was not possible because Aβ_x-40 was absent in both the control group and the group with DLB (no variance within groups).

Overall, the Aβ_x-40 burden (percentage of cortex with Aβ_x-40 immunoreactivity) paralleled Aβ_x-40 density data, with Aβ_x-40 densities being less in controls and patients with DLB than in patients with AD. These differences reached significance in all regions except the cerebellum.

The Aβ_x-42 plaque densities showed overall differences among controls (Figure 2, B), patients with DLB (Figure 2, D), and patients with AD (Figure 2, F), with the groups with DLB and AD showing greater densities of Aβ_x-42 plaques than controls. In the frontal gyrus, CA3, and PHG, significant differences were observed among the 3 groups, with the group with DLB having Aβ_x-42 plaque densities intermediate between those of the group with AD and controls. In the CA1 region, the control group had significantly fewer plaques than either of the other 2 groups. Differences between patients with AD and those with DLB were not significant in CA1. In the cerebellum, the group with AD showed greater densities than the group with DLB or the control group.

The Aβ_x-42 burden also showed overall differences among the 3 groups. The control group showed a lower burden than the group with DLB, and the group with DLB showed less Aβ_x-42 than the group with AD in all regions except the cerebellum, where the group with AD had greater Aβ_x-42 burden than either of the other groups.

The Aβ_x-40:Aβ_x-42 ratio in all regions was significantly greater in the group with AD than in either the group with DLB or controls, whether density or amyloid burden was assessed (Figure 3 and Table 3). The Aβ_x-40:Aβ_x-42 ratio of the group with DLB was sometimes less than that of the control group, although these differences did not reach significance because of the small amount of Aβ_x-40 in each group. In the cerebellum, ratios were 1.0 for the group with DLB and the control group and approached 0 for the group with AD because Aβ_x-40 plaques are exceedingly rare in the cerebellum, even in persons with advanced AD.

In most regions, patients with DLB (n=9) and those with AD (n=6) with the APOE ϵ4 allele had similar densities and percentage of area with AβP deposition compared with those lacking this allele (n=9 and n=4, respectively). In the control group, only 1 subject had the APOE ϵ4 allele. In patients with DLB, differences between groups with and without APOE ϵ4 varied, reaching significance in the PHG, where Aβ_x-40 densities were 23.7/mm² and 58.7/mm², respectively (t₁=2.809, P=.01). In patients with AD, the APOE genotype influenced Aβ_x-40 deposition more consistently, with mean Aβ_x-40 densities consistently greater in the group with AD with the ϵ4 allele than in those without the ϵ4 allele. These differences in Aβ_x-40 deposition reached significance in sections from the CA1 and the PHG (4.86/mm² and 16.97/mm² [t=2.726, P=.03] vs 1.54/mm² and 18.68/mm² [t=2.377, P=.04], respectively). The presence of Aβ_x-42 was not associated with increased AβP deposition.

We confirmed previous reports that patients with DLB, those with AD, and control subjects have AβPs. We extend this finding by determining that the pattern of AβP deposition differs between those with DLB and AD in all brain regions examined. In patients with DLB, Aβ_x-40 plaque densities were less than the densities in patients with AD. The Aβ_x-42 plaque densities were nearly as great in patients with DLB as they were in those with AD. Therefore, the Aβ_x-40:Aβ_x-42 ratio was less in patients with DLB than that in patients with AD. Also, the specimens of brain from patients with DLB lacking significant neuritic degeneration and severe neuronal losses contain increased densities of AβPs compared with age-matched controls. The pattern of AβP deposition, however, is similar in normal aging and in patients with DLB, with almost all AβPs being composed of Aβ_x-42; Aβ_x-40 plaques were nearly nonexistent.

Why the Aβ_x-40:Aβ_x-42 ratio in patients with DLB differs from that in patients with AD is unknown. Our observation that Aβ_x-40 densities in patients with DLB are less than those in patients with AD suggests that differences may occur between these 2 conditions in APP metabolism. We found no evidence to support the notion, however, that regional differences exist in the susceptibility to AβP formation in these 2 conditions because the regional distribution of AβP of both subtypes was similar in patients with AD and those with DLB. For example, in both conditions, the lowest density of AβPs (of either subtype) was found in the cerebellum, followed by the CA hippocampal subregions. The largest number of both Aβ_x-40 and Aβ_x-42 plaques was found in the cerebral cortex in patients with DLB and those with AD. Furthermore, the differences in Aβ_x-40:Aβ_x-42 ratios between patients with AD and those with DLB are not limited to specific brain regions.

The APOE genotype affects the likelihood of AD developing. Persons carrying the APOE ϵ4 allele are more likely to acquire AD, and symptoms are more likely to develop at an earlier age^34,35 than in those lacking this allele. In patients with AD, APOE affects the deposition of Aβ_x-40 and not Aβ_x-42.³⁶ Our data from patients with AD confirm this trend. In our study, however, the effect of the APOE genotype was less marked in patients with DLB than in those with AD. The overall poor relationship between the APOE genotype and Aβ in individual patients with DLB may be due to the paucity of Aβ_x-40 plaques in those with DLB. When group trends are examined, however, the frequency of the ϵ4 allele and the number of Aβ_x-40 plaques are lower in the group with DLB than in the group with AD. Therefore, our data do not contradict the hypothesis that APOE principally affects Aβ_x-40.

The relationship between in vivo neurotoxicity and Aβ subtypes remains unknown. We postulate that the subtype of Aβ deposited may be related to neuronal cell death. Many investigators think that Aβ_x-42 is the more toxic and aggressive form of Aβ because brain tissue from patients with early-onset AD (from both presenilin 1 and APP mutations) shows a reduced ratio of Aβ_x-40: Aβ_x-42.^13-15 Patients with AD who have both presenilin 1 and APP have more severe neuronal losses and at an earlier age than comparable patients with late-onset AD.^24,25,33 Our data are more compatible with the hypothesis that Aβ_x-42 is benign and that Aβ_x-40 (or the intraneuronal processes that contribute to Aβ_x-40 formation) is more strongly associated with neuronal degeneration. A previous study²⁶ reported that cortical neuronal loss is small in patients with DLB compared with the severe losses seen in those with AD. If Aβ_x-42 exerts a major, direct neurotoxic effect, we would expect to find much lower levels of Aβ_x-42 in the brain specimens of patients with DLB. Our immunohistochemical data from patients with DLB suggest that neuronal degeneration may occur after a threshold of Aβ_x-40 is reached. Any effect from differences in the aminoterminal of Aβ cannot be addressed by this study because the antibodies we used were specific only for the carboxyterminal.

The factors determining the relative proportion of Aβ_x-40 to Aβ_x-42 formed at the time of the metabolic breakdown of APP are not well understood. The subtypes Aβ_x-40 and Aβ_x-42 are both produced in the neuron during the metabolic degradation of APP.^16,17 Cell culture studies by Hartmann et al¹⁶ and Cook et al¹⁷ suggest that Aβ_x-42 synthesis occurs in the endoplasmic reticulum, whereas Aβ_x-40 is produced at a more distal point in the Golgi apparatus. Further characterization of these intraneuronal events will be important for our understanding of the role of Aβ in neuronal degeneration.

Recently, endoplasmic reticulum–associated Aβ protein (ERAB) has been shown³⁷ to bind Aβ_x-42 in patients with AD. This protein may be crucial to the pathogenesis of AD because ERAB is overexpressed in the brain of patients with AD. In addition, cell culture studies show that the toxic effect of Aβ on neurons is reduced when ERAB is blocked and increased when ERAB is overexpressed. The levels of ERAB in patients with DLB have not been determined. Because increased ERAB levels are associated with increased Aβ neurotoxicity, however, ERAB may not be increased in patients with DLB as neuronal losses in such patients are small.

Dickson et al²³ have termed patients with DLB with diffuse AβP deposition as having pathological aging. Although the patients with DLB in this study do not meet consensus criteria³⁰ for a high likelihood that their symptoms were due to AD, it could be argued that the increased amount of Aβ indicates that they are in an early stage of the AD process and thus had incipient AD. Our data cannot determine whether Aβ deposition is fundamentally different between patients with AD and those with DLB or whether plaque formation is in an earlier stage of AD, particularly because our patients with DLB had slightly shorter disease durations than the group with AD. Another study³⁸ of AβP deposition in patients with DLB that examined frontal cortex showed a similar trend. The near absence, however, of Aβ_x-40 in patients with Aβ_x-42 densities approaching those of patients with AD suggests that the events leading to Aβ deposition in patients with AD and those with DLB may differ. Further study of factors determining which APP metabolite is formed in patients with DLB, those with AD, and as a result of normal aging may advance our understanding of the natural history of APP and Aβ in neurodegenerative diseases.

Accepted for publication December 11, 1998.

This study was supported in part by grant 13623 from the National Institute on Aging, Bethesda, Md, and the Robert Potamkin Fund (Dr Lippa). We acknowledge Brendan O'Connell for assistance with data acquisition, the Joseph and Kathleen Bryan Brain Bank at Duke University Medical Center, Durham, NC, which is supported by grant AG05128 from the National Institute on Aging and Glaxo Wellcome Inc, Research Triangle Park, NJ. We also acknowledge the Grant in Aid for Scientific Research on Priority Area (Dr Mori).

Reprints: Carol F. Lippa, MD, Department of Neurology, MCP-Hahnemann University, 3300 Henry Ave, Philadelphia, PA 19129 (e-mail: lippa@auhs.edu).