Olfactory Bulb Amyloid-β Correlates With Brain Thal Amyloid Phase and Severity of Cognitive Impairment

Abstract

The Alzheimer disease (AD) neuropathological hallmarks amyloid β (Aβ) and tau neurofibrillary (NF) pathology have been reported in the olfactory bulb (OB) in aging and in different neurodegenerative diseases, which coincides with frequently reported olfactory dysfunction in these conditions. To better understand when the OB is affected in relation to the hierarchical progression of Aβ throughout the brain and whether OB pathology might be an indicator of AD severity, we assessed the presence of OB Aβ and tau NF pathology in an autopsy cohort of 158 non demented control and 173 AD dementia cases. OB Aβ was found in less than 5% of cases in lower Thal phases 0 and 1, in 20% of cases in phase 2, in 60% of cases in phase 3 and in more than 80% of cases in higher Thal phases 4 and 5. OB Aβ and tau pathology significantly predicted a Thal phase greater than 3, a Braak NF stage greater than 4, and an MMSE score lower than 24. While OB tau pathology is almost universal in the elderly and therefore is not a good predictor of AD severity, OB Aβ pathology coincides with clinically-manifest AD and might prove to be a useful biomarker of the extent of brain spread of both amyloid and tau pathology.

INTRODUCTION

The presence of Alzheimer disease (AD) neuropathological hallmarks, such as amyloid β (Aβ) pathology and tau neurofibrillary changes (NF), was reported in olfactory relevant regions, particularly the olfactory bulb (OB), in aging and in different neurodegenerative diseases at postmortem examination. The OB is of particular interest in neurodegenerative diseases because it is affected in early disease stages and has been suggested to be an entry point for potential pathogens to enter and then spread throughout the brain (1–3).

Indeed, we and others have previously reported on the presence in the OB of tau and α-synuclein (αSyn) histopathology at early stages of AD, Lewy body disease (LBD), Parkinson disease (PD) and multiple system atrophy (MSA), as well as the presence of other specific molecular markers in amyotrophic lateral sclerosis, progressive supranuclear palsy, corticobasal degeneration, Pick disease, and Huntington disease (1, 4–17) These changes are consistent with olfactory dysfunction, which has been repeatedly reported to manifest early in neurodegenerative diseases (18). The changes provide a potential model of olfactory-mediated pathogenesis in neurodegenerative disorders, including AD and LBD, and suggest the potential for new olfactory-related diagnostic modalities.

While tau NF pathology was found to be almost universal in older nondemented individuals, the presence of Aβ plaques has been reported less frequently and found to occur in cases with more severe AD pathology suggesting that the presence of Aβ pathology in the OB could be an indicator of the diagnostic levels of AD (4–6, 8, 9, 19, 20). The hierarchical progression of Aβ pathology throughout the brain was described in 5 Thal amyloid phases in which the pathology typically progresses from the neocortex to limbic regions, basal ganglia, brainstem, and finally to the cerebellum (21). Nevertheless, involvement of the OB has not been specifically assessed in relation to these Thal phases. It has previously been demonstrated that Aβ in the OB correlated with whole-brain Thal amyloid phases and Braak NF stages (6). We sought to extend this work by determining the Thal phase at which OB Aβ deposits are most likely to first appear, and how well the presence and densities of OB tau and Aβ deposits might be suitable indicators of AD stage, histopathological severity, and severity of cognitive impairment.

Therefore, this clinicopathological study aimed to investigate the severity of Aβ and tau deposition in the OB in relation to the progression of whole-brain AD pathology and cognitive impairment. We determined the densities of OB Aβ and tau NF pathologies in a large cohort of 331 autopsied cases that were neuropathologically distributed along a continuum of histopathological AD stages using tissue and data from the Arizona Study of Aging and Neurodegenerative Disorders (AZSAND) and Banner Sun Health Research Institute Brain and Body Donation Program (BBDP) (22).

MATERIALS AND METHODS

Database and Subject Selection

All subjects included in this study were selected from the AZSAND/BBDP database (www.brainandbodydonationprogram.org) (22), and all subjects signed informed consents that were ethically approved by designated Banner Sun Health Research Institute Institutional Review Boards. Subjects are all volunteers and most receive standardized research-dedicated clinical assessments including an extensive neuropsychological test battery and cognitive and movement disorders assessments by board-certified subspecialty neurologists. Additional clinical information is periodically gathered from requisitioned private medical records. Test results available included the Mini-Mental State Exam (MMSE) and the Unified Parkinson’s Disease Rating Scale (UPDRS).

At the time of death, a full neuropathological examination is performed, including assignment of the AD Braak NF stage (23), the Thal amyloid phase for Aβ plaque brain distribution (21), the CERAD neuritic plaque density score (24), and αSyn stage according to the Unified Staging System for Lewy Body Disorders (USSLBD) (25). Data also include regional and summary cortical brain density measures for tau NF brain and plaque load (for both, there is a total possible score of 15 based on summary of 0–3 scores in each of 5 regions: frontal association cortex, parietal association cortex, temporal association cortex, hippocampus CA1, and entorhinal/transentorhinal areas). The regional and summary brain load of αSyn pathology is also recorded (total possible score of 40 based on summary of 0–4 scores for each of 10 brain regions).

Cases included were selected to cover the full range of AD pathology and were categorized at their final consensus clinical conference as either controls without dementia or parkinsonism (n = 158) or as AD dementia (n = 173), which was defined as having an “intermediate” or “high” rating on the NIA-Reagan system. Because many of the cases were autopsied prior to the publication of the newer NIA-AA rating system, the NIA-Reagan system was used for all subjects. Some AD subjects presented with additional comorbid neuropathologically diagnosed conditions, including PD (n = 10), DLB (n = 12), vascular dementia (n = 13), and progressive supranuclear palsy (n = 2).

Immunohistochemical Assessment of OB Pathology

For the purpose of this study, additional OB immunohistochemical staining was performed. Tissue samples of the OB were embedded in paraffin, sectioned at 5-μm thickness, and mounted on charged glass slides for histology. Immunohistochemical staining was then performed using the monoclonal antibody clone 6E10 (BioLegend, San Diego, CA) for specific detection of all types of Aβ pathology including neuritic, cored, and diffuse plaques. The primary antibody concentration was 1:1000. Staining for tau pathology was performed with the AT8 antibody to phosphorylated tau (Thermo Fisher Scientific, Waltham, MA), diluted to 1:1000. For both types of staining, the epitope exposure method consisted of 20 minutes in 80% formic acid. Secondary antibodies and signal development were performed with an immunoperoxidase method as previously described (26). The presence of pathology in the OB was semiquantitatively assessed by an experienced neuropathologist (T.G.B), who was blinded to clinical data. The density of Aβ plaques was scored from 0 to 3, as 0 for absent, 1 for sparse, 2 for moderate, and 3 for frequent, based on templates published by CERAD (24). The density of OB tau pathology was graded from 0 to 4, based on our previously published templates (27).

Photomicrographs depicting Aβ and tau pathologies in the OB are shown in Figure 1.

FIGURE 1.

Photomicrographs depicting (A) amyloid β and (B) tau immunohistochemical pathologic inclusions in the OB at low and high magnifications.

Because OB lesions were reported to most frequently and heavily affect the anterior olfactory nucleus (AON), which consists of multipolar medium-sized neurons arranged in small discontinuous groups along the central axis of the OB, we also noted the localization of the lesions in the AON and in the outer layers of the OB (altogether including glomerular, external plexiform, mitral cell, internal plexiform, and granule cell layers), in a subset of 60 cases for each stain (1).

Data Analysis

Statistical analysis was performed using SPSS software (IBM SPSS Statistics 23.0). The link between OB Aβ and tau pathology scores and other clinical and neuropathologic characteristics was assessed using non-parametric univariate Spearman correlations. Non-parametric Mann-Whitney U-test or chi-square tests, as appropriate, were used to compare these measures between sexes and for the differences between OB Aβ-positive cases (cases with the presence of Aβ in the OB) and OB-Aβ negative cases. The relationships between OB Aβ and tau pathology scores with whole-brain Thal amyloid phase, Braak NF stage, and last MMSE score were further investigated using logistic regression models adjusting for age and sex. Receiver operating characteristics (ROC) were analyzed to assess the ability of OB Aβ presence for discrimination between a lower (0–3) and a higher (4–5) Thal phase.

RESULTS

Table 1 depicts basic demographic, clinical, and neuropathological characteristics of the cases studied. The presence of Aβ in the OB was found in less than 5% of cases in lower Thal phases 1 and 2, in 60% of cases in phase 3 and in more than 80% in higher Thal phases 4 and 5. Tau pathology was found in more than 95% of cases and as early as Thal phase 0 (Fig. 2 and Table 2). In most cases, lesions were located in both the AON and the outer layers of the OB but the AON was most frequently and heavily involved. Aβ lesions were located in both the AON and outer layers of the OB in 75% of cases, only in the AON in 17% of cases and only in outer layers in 8% of cases. For tau pathology, they were in both regions in 87% of cases, only in the AON in 8% of cases and only in outer layers in 5% of cases. Localization of pathology was not related to severity of pathology in the bulb.

TABLE 1.

Demographic and Neuropathological Characteristics of Cases

	All Cases	ADD	Controls	OB Aβ Positive Cases	OB Aβ Negative Cases
Number of cases	331	173	158	174	157
Age	84.6 ± 7.8	82.7 ± 8.6	86.6 ± 6.7	84.1 ± 8.2	84.9 ± 7.7
Sex (M/F)	160/171	87/86	73/85	86/88	74/83
% Apo E genotype	39.4	42.9	35.6	40.8	37.7
MMSE	19.8 ± 10.2	11.6 ± 8.2	28.2±1.5	15.2 ± 9.9	24.97 ± 7.8*
UPDRS	15.4 ± 19.1	31.4 ± 25.0	8.1 ± 8.6	22.1 ± 22.8	10.6 ± 14.0*
OB Aβ density	1.4 ± 1.4	2.2 ± 1.2	0.6 ± 1.1	2.7 ± 0.6	0.0 ± 0.0*
OB tau density	2.2 ± 0.9	2.5 ± 0.8	1.9 ± 0.9	2.6 ± 0.7	1.7 ± 0.9*
Thal phase	3.0 ± 1.8	4.2 ± 1.0	1.7 ± 1.8	4.1 ± 1.1	1.8 ± 1.7*
Braak stage	4.0 ± 1.3	4.7 ± 1.1	3.2 ± 0.9	4.5 ± 1.1	3.4 ± 1.3*
USSLBD stage	1.1 ± 1.4	1.6 ± 1.5	0.6 ± 1.0	1.4 ± 15	0.7 ± 1.2*
NF brain load	7.9 ± 4.5	10.6 ± 4.2	5.0 ± 2.4	9.8 ± 4.2	5.8 ± 3.7*
Plaque brain load	9.0 ± 5.5	12.7 ± 2.1	5.0 ± 5.3	12.5 ± 2.4	5.1 ± 5.4*
αSyn brain load	5.6 ± 9.3	8.9 ± 10.9	2.5 ± 5.9	7.4 ± 10.27	3.8 ± 7.7*

Data are presented as mean ± SDs.

ADD, Alzheimer disease dementia; ApoE, apolipoprotein E genotype; MMSE, last Mini Mental State Examination score; UPDRS, last Unified Parkinson’s Disease Rating Scale motor score (part 3 motor score); USSLBD stage, Unified staging system for Lewy body disorders; Plaque Score and NF Score, summary regional brain density of neurofibrillary tangles and plaques scores with maximum scores of 15; αSyn Score, summary regional brain Lewy-type synucleinopathy density score with a maximum score of 40.

^*Significant differences between OB-Aβ positive cases and OB Aβ-negative case.

FIGURE 2.

Percentage of cases with olfactory bulb Amyloid β immunoreactivity for each Thal Phase.

TABLE 2.

Cases with Olfactory Bulb Amyloid β and Tau Pathology for Thal Phase and Braak NF Stage

(a) Thal Phase
	Thal Phase 0	Thal Phase 1	Thal Phase 2	Thal Phase 3	Thal Phase 4	Thal Phase 5
OB Aβ+ cases (%)	2 (3.5)	1 (5.9)	9 (21.4)	47 (61.8)	25 (92.6)	90 (80.4)
OB tau+ cases (%)	53 (93.0)	15 (88.2)	41 (97.6)	75 (98.7)	27 (100)	112 (100)
Total number of cases	57	17	42	76	27	112

(b) Braak NF Stage
	Braak NF 1	Braak NF 2	Braak NF 3	Braak NF 4	Braak NF 5	Braak NF 6
OB Aβ+ cases (%)	0 (0)	6 (28.5)	25 (31.6)	54 (51.5)	56 (86.2)	33 (70.2)
OB tau+ cases (%)	10 (71.4)	21 (100)	78 (98.7)	102 (97.1)	65 (100)	47 (100)
Total number of cases	14	21	79	105	65	47

OB Aβ-positive and -negative cases did not differ in age (p = 0.3) or sex (χ² = 0.1735; p = 0.68), but positive cases had significantly lower last MMSE score; and higher last motor UPDRS scores, OB tau pathology score, Thal phase, Braak NF stage and total brain plaque, tangle and αSyn brain scores (all p < 0.001). Mean age (U = 10292.5, p < 0.001) and Braak stage (U = 11617.0, p = 0.015) were higher overall in females, while no sex differences were observed for OB tau density, Aβ density, Thal phase, or MMSE score (Table 1).

OB Aβ density scores were found to correlate with Thal phases (Rho = 0.616, p < 0.001), Braak NF stages (Rho = 0.449, p < 0.001), total brain plaque score (Rho = 0.641, p < 0.001), total tangles brain score (Rho = 0.491, p < 0.001), and total αSyn brain score (Rho = 0.198, p < 0.001), as well as USSLBD stage (Rho = 0.182, p = 0.001), MMSE score (Rho = −0.534, p < 0.001), and UPDRS motor score (Rho = 0.332, p < 0.001) but did not correlate with age (Rho = −0.035, p = 0.4; Fig. 3).

FIGURE 3.

Correlation between olfactory bulb Amyloid β and MMSE, Thal amyloid phase and Braak NF stage.

OB tau density scores were found to significantly correlate with Thal phases (Rho = 0.439, p < 0.001), Braak NF stages (Rho = 0.511, p < 0.001), total brain plaque score (Rho = 0.482, p < 0.001), total brain tangle score (Rho = 0.550, p < 0.001), and total brain αSyn score (Rho = 0.196, p = 0.001), as well as USSLBD stage (Rho = 0.161, p = 0.003), MMSE score (Rho = −0.360, p < 0.001), UPDRS motor score (Rho = 0.250, p < 0.001), and age (Rho = 0.113, p = 0.04).

In logistic regression models, OB pathology predicted a higher Thal phase than 3 [χ²(4) = 116.910; p < 0.001; R² = 0.298], with covariates for sex and age, significant independent predictors were OB Aβ, OB tau, and age. A higher Braak stage than 4 could significantly be predicted [χ²(4) = 109.535; p < 0.001; R² = 0.282] with significant independent predictors being OB Aβ, OB tau, age, and sex. A lower MMSE score than 24 was also significantly predicted [χ²(4) = 118.990; p < 0.001; R² = 0.315] with significant independent predictors being OB Aβ, OB tau, and age. Significance, odds ratios, and 95% confidence intervals of each significant independent predictor as well as sensitivity and specificity of whole models are shown in Table 3.

TABLE 3.

Logistic Regression Analyses With Thal Phase, Braak Stage, and Last MMSE Score as the Dependent Variable

Predictors	p-Value	Odds Ratios	95% CI	Specificity	Sensitivity
(a) Predicting a higher Thal phase than 3
Equation	<0.001			78.6	73.3
OB Aβ	<0.001	2.139	1.715, 2. 667
OB tau	0.02	1.846	1.259, 2.708
Age	0.01	0.942	0.909, 0.977
Sex	0.2
(b) Predicting a higher Braak stage than 4
Equation	<0.001			82.8	63.1
OB Aβ	<0.001	1.683	1.355, 2.090
OB tau	<0.001	2.592	1.717, 3.912
Age	<0.001	0.909	0.873, 0.945
Sex	0.02	1.979	1.112, 3.520
(c) Predicting a lower MMSE score than 24
Equation	<0.001			78.2	78.5
OB Aβ	<0.001	2.167	1.746, 2.691
OB tau	0.02	1.543	1.355, 2.090
Age	<0.001	0.917	0.881, 0.955
Sex	0.2	1.447	0.826, 2.533

Significant independent predictors in regression models are presented in bold.

To better assess how specifically OB Aβ can predict Thal phases, we computed analysis of ROC curves that showed that Aβ presence in the OB can predict a higher Thal phase than 3 with a sensitivity of 80.2%, a specificity of 63.3%, and an area under the curve of 0.717 (Fig. 4).

FIGURE 4.

ROC curve for the prediction of a higher whole-brain Thal phase greater than 3 by presence or absence of OB Aβ. Sensitivity is 80.2% and specificity is 63.3% with an area under the curve of 0.717.

DISCUSSION

This clinicopathological study evaluates the link between the presence of Aβ and tau pathologies in the OB and the severity of AD pathology in the brain. We show that OB Aβ inclusions, though rare in low amyloid Thal phases, are present in two-thirds of cases in Thal phase 3 and in more than 80% of cases in higher Thal phases 4 and 5. OB Aβ and tau pathology both significantly predicted a Thal phase greater than 3, a Braak NF stage greater than 4, and an MMSE score lower than 24. While tau pathology is almost universal in OBs of older individuals and therefore not useful to stage AD severity, OB Aβ pathology may be a better predictor of diagnostic levels of AD neuropathology and cognitive decline.

This is the first study to specifically assess the appearance of Aβ pathology in the OB in relation to the topographical whole-brain progression of Aβ deposition according to the hierarchical sequence described by Thal and colleagues (21, 28). We report the presence of OB in about 60% of cases in amyloid Thal phase 3 and in more than 80% of cases in higher Thal phases 4 and 5. With regards to Braak NF stages, 50% of cases with Braak NF stage 4 presented OB Aβ pathology and up to 80% of cases in Braak NF stages 5 or 6 presented OB Aβ pathology. Even though the OB is typically suggested to be one of the first regions affected in neurodegenerative diseases and was shown to have tau and αSyn pathology in very early stages of disease, this study along with previous work suggests that Aβ pathology affects the OB only in more advanced AD disease stages (5, 6, 8, 11). Exact mechanisms of Aβ initial deposition or progression are not fully understood, hence it is not yet clear as to why brain regions are involved in a particular sequence (29). It has been demonstrated, however, that the neocortex is first affected, in phase 1, and that later regions are subsequently affected based on their neuroanatomical connections with the neocortex (21, 28, 29). The OB is heavily connected with the amygdala (30) and it is therefore possible that Aβ “spreads” from the amygdala to the OB. According to Thal phases, the amygdala is affected in 88% of cases in Thal phase 3 and in up to 100% of cases in Thal phase 4 while the OB has Aβ deposits in about 60% of Phase 3 cases and 80% of phases 4 and 5. Alternatively, brain regions may simply be differentially affected based on their inherent selective vulnerability.

We found that both Aβ and tau OB pathologies correlated with the severity of whole-brain AD pathology as well as cognitive impairment as measured by the last MMSE. Logistic regression analysis demonstrated that both OB Aβ and tau pathologies significantly predict a higher Thal phase than 3, a higher Braak stage than 4 and a lower MMSE score than 24, when adjusted for age and sex. Odds ratios were higher for OB Aβ than for OB tau pathology to predict a higher Thal phase and a lower MMSE score. Further, independently of tau, the presence of Aβ in the OB can predict a higher Thal phase than 3 with a sensitivity of 80.2%, a specificity of 63.3%. These results support that the appearance of OB Aβ may roughly correspond to the presence of cognitive symptoms and dementia in subjects (21, 31–34). Even though some studies reported that amyloid Thal phase does not appear to be an independent predictor of cognitive changes (35) other demonstrated its linkage with clinical diagnosis and as a predictor of dementia (36). The almost universal presence of tau pathology in OBs of elderly humans with or without AD, appearing as early as Thal phase 0 and Braak NF stage 1, may reduce its usefulness as a binomial marker of disease presence or absence although its density still correlates with brain disease stage severity. Altogether, these results indicate that OB Aβ may prove more useful as a biomarker of the extent of AD disease severity.

Potential techniques that assess the level of Aβ OB pathology may help to clinically confirm a high level of AD pathology in the brain. Thus, tests of olfaction (18, 37–48), as well as OB biopsies or imaging (1, 6, 49–51) or nasal mucosal sampling (52–61) might be diagnostically useful with the additional advantage of also indicating the presence of comorbid neurodegenerative pathologies including LBD, MSA, non-AD tauopathies, and TDP-43. While OB biopsy may be too invasive (50) and would require specialized surgical personnel, the close proximity of primary olfactory sensory neurons in the olfactory mucosa provides a less-invasive substrate, through RT-QuIC or analogous methodology for detecting seed-induced protein aggregation in nasal brushings. This has been successfully demonstrated for prion disease (62–65), and, in preliminary work, for LBD (56, 57, 66, 67). In the future, these highly sensitive methods may also allow minimally invasive fine needle aspiration biopsy of the OB through the existing foramina of the cribriform plate (68). Future studies may also investigate enhanced analyses of olfactory function that may also provide additional disease-stage profiling (38).

We acknowledge some limitations to this study. First, only those cases with an AD clinicopathological diagnosis or control status were evaluated. Future studies should also investigate other neurodegenerative diseases. Moreover, we did not have access to premortem olfactory function of the autopsied cases and therefore could not investigate the link between olfactory performance and the presence of OB Aβ and tau pathologies.

In conclusion, results from this large neuropathological investigation demonstrate that Aβ deposition in the OB correlates with Thal amyloid phase and is observed in two-thirds of subjects in Thal phase 3 and in most cases in Thal phases 4 and 5. Further, OB Aβ and tau pathologies predict diagnostic levels of AD and severity of cognitive impairment emphasizing the promise of olfactory-related diagnostic modalities for staging of AD and potentially other related neurodegenerative diseases.

 

The Brain and Body Donation Program has been supported by the National Institute of Neurological Disorders and Stroke (U24 NS072026 National Brain and Tissue Resource for Parkinson’s Disease and Related Disorders), the National Institute on Aging (P30 AG19610 and P30AG072980, Arizona Alzheimer’s Disease Center), the Arizona Department of Health Services (contract 211002, Arizona Alzheimer’s Research Center), the Arizona Biomedical Research Commission (contracts 4001, 0011, 05-901, and 1001 to the Arizona Parkinson’s Disease Consortium), and the Michael J. Fox Foundation for Parkinson’s Research.

The authors have no duality or conflicts of interest to declare.