Olfactory Dysfunction and Incidence of Motoric Cognitive Risk Syndrome

Abstract

Background and Objectives

To examine associations between olfactory dysfunction, Alzheimer disease (AD) pathology, and motoric cognitive risk syndrome (MCR), a predementia syndrome characterized by cognitive complaints and slow gait that is associated with risk for AD and other dementias.

Methods

We conducted a retrospective analysis of a prospective cohort study to examine whether baseline olfactory function was associated with the risk of incident MCR in 1,119 adults aged 60 years and older (75.1% female). The association between performance on the Brief Smell Identification Test (BSIT) and incident MCR risk was computed using Cox models and reported as the hazard ratio (HR) with 95% CIs adjusted for demographic, comorbidity, and cognitive factors. Furthermore, we assessed the relationship between postmortem AD pathology and non-AD pathology and olfactory function at the time of MCR diagnosis using linear regression models adjusted for sex, education, age at death, and time from diagnosis to death.

Results

There were 544 (48.6%) incident cases of MCR over a median follow-up of 3.94 years. Lower BSIT scores (poor olfaction) at baseline were associated with an increased risk of incident MCR (HR for a 1-point increase in BSIT score 0.92; 95% CI 0.88–0.96) in fully adjusted models. Those with hyposmia (scores of ≤8 on the BSIT) at baseline (26.6%) were at an increased risk of MCR (HR 1.44; 95% CI 1.19–1.74) compared with those with normal olfactory function. Higher levels of the composite measure of global AD pathology and presence of Lewy body pathology were associated with lower BSIT scores at the time of incident MCR diagnosis (n = 118). τ tangle density, a specific component of AD pathology, was inversely associated with olfactory function, and the correlation remained after controlling for mild cognitive impairment syndrome and the presence of Lewy body pathology.

Discussion

The results provide evidence that olfactory dysfunction precedes MCR incidence and is related to Alzheimer pathology, providing a clinical approach to risk stratify and subtype MCR.

The motoric cognitive risk syndrome (MCR) is a predementia syndrome characterized by subjective cognitive complaints and slow gait speed.¹ MCR was established as studies have demonstrated that impaired gait is an early feature of the dementia disease process.^2,3 MCR is common in older adults and is a strong predictor of major cognitive decline and dementia, of both the Alzheimer disease (AD) type⁴ and vascular type.⁵ MCR is associated with an increased dementia risk independent of mild cognitive impairment (MCI).⁴ Although there has been significant progress in establishing the epidemiology of MCR, features of prodromal stages of MCR and associated risk factors remain to be established.

Impaired olfactory function is associated with cognitive decline in older patients.^6,7 Furthermore, difficulty in familiar odor identification was reported to predict MCI and risk of conversion to AD.⁸ Hyposmia is correlated with Alzheimer pathology, namely β-amyloid and neurofibrillary tangles.^9,10 Neurofibrillary tangles, a key pathologic feature of AD,¹¹ are found throughout olfactory neurons in the olfactory bulb in patients with AD, and antemortem odor identification deficits correlate with postmortem tangles in the olfactory bulb and its projection targets.^9,12,13 Amyloid plaques are also detected in the olfactory bulb and projection sites in patients with AD, yet to a lesser extent than tangles.^9,12 Olfactory dysfunction also precedes Lewy body disease, and its associated α-synuclein pathology involves olfactory centers very early in the disease course.^14,15 There are mixed findings on the association between vascular pathology and olfactory dysfunction. Some studies have shown clinical stroke as a risk factor for impaired olfaction,¹⁶ whereas others have shown no association.^17,18 There are limited data on associations between olfaction and other cerebrovascular pathologies.

Olfactory function in older adults is associated with gait speed, mobility, and balance in older adults, even when controlling for cognitive function.¹⁹ Olfactory dysfunction is also associated with frailty,²⁰ and frailty in turn is associated with an increased risk for developing MCR, irrespective of MCI.²¹ The relationship between olfactory dysfunction and gait may arise from the projection of olfactory tracts to the orbitofrontal cortex and cerebellum²² because these brain structures are involved in mobility, movement planning, spatial navigation, and sensorimotor integration.²³ Although these associations have been demonstrated, the longitudinal association between impaired impaired olfaction and risk of MCR is unknown. Also, the relationship between olfactory dysfunction and dementia-related pathology in those diagnosed with MCR is not established.

We hypothesized that impaired olfactory function may be associated with an increased risk of incident MCR. Furthermore, we hypothesized that olfactory function in people with MCR would be correlated with higher levels of AD pathology, specifically neurofibrillary tangles. To test these hypotheses, we analyzed participant data collected by the Rush Memory and Aging Project (MAP). Elucidating the role of olfactory function in MCR risk and pathology may provide insight into the biological pathways of MCR and potential therapeutic strategies to target for dementia treatments.

Methods

Study Cohort

This study includes data from a subset of 2,179 total participants enrolled in the Rush MAP, an established longitudinal clinical-pathologic study of chronic conditions of aging. Study goals and design have been reported.^24,25 Participants were recruited from retirement communities around northeastern Illinois. Eligible participants agreed to organ donation and annual detailed clinical evaluations conducted at participants' homes.²⁴

MCR Diagnosis

MCR diagnosis is adapted from the criteria for MCI,⁴ replacing the objective impairment on cognitive tests with slow gait. MCR is defined by the presence of both subjective cognitive complaints and slow gait in older ambulatory individuals without dementia as previously reported.⁴ Cognitive complaints were identified if participants answered very often, often, and sometimes to either question “do you have trouble remembering?” or “is your memory worse than 10 years ago?” Gait was timed for each participant over a fixed distance of 8 feet (2.4 m) at their normal pace from a standing start point and converted to speed (centimeters per seccond).²⁴ Slow gait was defined as walking speed 1 SD below age- and sex-specific means. Gait speed measured using timed methods have demonstrated strong reliability.²⁶

Olfactory Function

Participants were assessed on their ability to identify familiar odors at baseline and at subsequent follow-ups using the 12-item Brief Smell Identification Test (BSIT). Twelve microcapsules, each containing a different familiar odor, were scratched and placed under the nose of the participant, who then attempted to match the smell to 1 of 4 choices. Scores were calculated as the number of odors correctly identified out of 12, with a score of 0.25 assigned to missing answers to a maximum of 2 allowed. Strong test validity measures have been reported.⁸ Scores were examined continuously and dichotomously, with a score of ≤8 considered hyposmia and scores >8 considered normal olfaction as previously described.^27,28

Covariates

Demographic characteristics included age, sex (male or female), education years, and race/ethnicity. Race/ethnicity was assessed in response to the questions “with which group do you most closely identify yourself?” (responses [choose 1]: White; Black, Negro, African-American; Native American, Indian; Eskimo; Aleut; Asian or Pacific Island) and “are you of Spanish/Hispanic/Latin origin?” (response: yes or no). Covariates were selected based on previously reported associations with MCR or olfaction. Comorbidities were included as composite measure of 7 common medical conditions present in at least 5% of participants at baseline. Hypertension, diabetes mellitus, heart disease, cancer, thyroid disease, and head injury with loss of consciousness were based on self-report of a physician diagnosis. A diagnosis of stroke was based on the history plus neurologic examination. Depressive symptoms were quantified using the 10-item Center for Epidemiologic Studies–Depression Scale²⁹ because previous studies have indicated an association with incident MCR.⁵ Smoking habits were defined as never smoked, former smoker, or current smoker.³⁰ The body mass index (BMI) was calculated from height and weight. Obesity was defined as a BMI ≥30 kg/m².²⁵ The probability of parkinsonism was diagnosed by clinicians as previously described²⁴ and dichotomized as not present or possible to highly probable. Mobility impairment was assessed using the Rosow-Breslau scale (higher scores indicating worse),³¹ which measures the ability to do 3 activities: doing heavy work around the house, walking up and down stairs, and walking half a mile without help. Lifetime daily alcohol intake was a log-transformed measure of the number of alcoholic drinks consumed per day during the period that participants drank the most in their life. A composite semantic memory score was calculated as previously described²⁵ and assessed using a 15-item version of the Boston Naming Test, a 15-item reading test, and Verbal Fluency test. A composite global cognitive function score comprised 19 tests including the semantic memory tests and tests of episodic memory, working memory, perceptual speed, and visuospatial ability.³² The individual cognitive tests, composite scores, and diagnosis of dementia and MCI have been detailed elsewhere.^24,25,32

Pathology

We examined the relationship between olfactory function at the time of incident MCR diagnosis and postmortem AD and non-AD pathology. The BSIT score at the time of the first incident MCR diagnosis was used to control for progression of cognitive decline in MCR from initial diagnosis to dementia. Autopsy procedures in MAP have been described previously.⁹

First, 5 brain regions (midfrontal gyrus, inferior parietal gyrus, middle temporal gyrus, entorhinal cortex, and hippocampus [CA1/subiculum]) of interest were stained with Bielschowsky silver, and a neuropathologist blinded to all clinical data separately counted neuritic plaques, diffuse plaques, and neurofibrillary tangles. Standardized scores from each region were averaged to yield composite measures. Global AD pathology was calculated using an average of these 3 measures.⁹

A more systematic, impartial, and molecule-specific means of quantifying AD pathology was achieved using immunohistochemistry and image analyses. Eight brain regions (hippocampus, entorhinal cortex, midfrontal cortex, inferior temporal, angular gyrus, calcarine cortex, anterior cingulate cortex, and superior frontal cortex) were examined for fraction of area containing β-amyloid immunoreactive plaques and density of paired helical filament (PHF) τ tangles, and then each averaged to yield composite measures. Further information on procedures to determine pathology has been previously reported.⁹

Lewy body pathology was examined as 4 stages of distribution of α-synuclein in the brain using immunostaining techniques.³³ Lewy body disease was categorized as not present, predominantly nigral Lewy bodies, limbic Lewy bodies, or neocortical Lewy bodies. For this study, Lewy body pathology was dichotomized as present or absent. Cerebrovascular pathologies examined included gross infarcts, microinfarcts, cerebral amyloid angiopathy, atherosclerosis, and arteriolosclerosis as previously described.³⁴ Vascular pathology was examined as a composite variable defined as the presence or absence of any one of the cerebrovascular pathologies.

Statistical Analyses

Baseline characteristics for the overall sample were examined with descriptive statistics. Baseline characteristics of participants who did and did not develop MCR were compared using the independent samples t test for normally distributed continuous variables, the Mann-Whitney U test for not normally distributed continuous variables, and the Pearson χ² test for categorical variables.

To examine the association between baseline olfaction and incident MCR, Cox proportional hazards models were used to compute the hazard ratios (HRs) and 95% CIs. Time to event was calculated in years from baseline to the first visit at which MCR was diagnosed or to final study contact, whichever came first. The eligible sample did not include individuals younger than 60 years at baseline, nor those missing assessments for olfaction, gait speed, or cognitive complaints at baseline. Participants with dementia or MCR at baseline were also excluded. The first model controlled for baseline age, sex, education, comorbidities score, and depression symptoms. Linking an odor to a label of its name requires intact semantic memory³⁵; therefore, we repeated the analysis including the composite semantic memory score. In addition, obesity and smoking have been shown to be associated with both impaired odor identification^16,36 and MCR.^5,37 Thus, the second model incorporated terms to control for obesity, smoking status, and semantic memory in addition to those in model 1. A third model including covariates from models 1 and 2 as well as the baseline presence of parkinsonism, mobility impairment, and lifetime daily alcohol intake was examined. To establish a clinically relevant marker, we categorized BSIT scores ≤8 to define those with hyposmia as previously described.^27,28 Proportional hazards assumptions of all models were examined analytically and graphically and were adequately met.

We conducted sensitivity analyses to account for overlap between MCR and MCI by excluding prevalent MCI and participants who developed incident MCI on follow-up visits at or before the visit at which incident MCR was diagnosed. To examine the possibility that olfaction declines very early in the transition between normal cognitive function and MCR, we excluded cases diagnosed with incident MCR within the first 3 years of follow-up. Three years was chosen to mirror the findings of a previous study, which demonstrates that MCR is an early marker of cognitive impairment.⁴

Linear regression models were used to examine the association between pathology measures and olfactory function at the time of incident MCR diagnosis in the subset of participants who underwent brain autopsy. First, we examined the association between global AD pathology and the BSIT score at the time of MCR diagnosis adjusting for potential confounding effects of age at death, sex, education, and time from MCR diagnosis to death. This model was then repeated for each of the subscores (neurofibrillary tangles, neuritic plaques, and diffuse plaques) as well as β-amyloid plaque deposition, tangle density, presence of Lewy body pathology, and presence of composite vascular pathology. To account for multiple comparisons of pathology factors, we applied Bonferroni correction and set an alpha level of 0.007.

To examine the independent association between types of pathology and olfaction, we incorporated global AD pathology and Lewy body pathology into the same model with adjustment for covariates described earlier. To account for overlap between MCR and MCI in global AD pathology burden and tau tangle density, we conducted sensitivity analyses excluding individuals with MCI at baseline and those who developed MCI at or before the visit when incident MCR was diagnosed.

Assumptions of all models were tested graphically and analytically by examining normality, linearity, homoscedasticity, and absence of multicollinearity and were adequately met. All analyses were conducted using SPSS version 27 (SPSS Inc., Chicago, IL).

Standard Protocol Approvals, Registrations, and Patient Consents

The MAP was approved by an Institutional Review Board of Rush University Medical Center. All participants signed written informed consent, an Anatomical Gift Act to donate their brain, and a repository consent to share data and biospecimens. The Institutional Review Board of the Albert Einstein College of Medicine approved this analysis.

Data Availability

Data are available via the Rush Alzheimer's Disease Center Research Resource Sharing Hub (radc.rush.edu).

Results

Study Population

Of the 2,179 participants who had completed baseline evaluations, we excluded 23 individuals younger than 60 years at baseline and those missing baseline assessments for olfaction (n = 280), gait speed (n = 128), and cognitive complaints (n = 3). We excluded 116 people with dementia and 365 with MCR at baseline. As we aimed to examine olfactory function as an early marker of MCR and not general cognitive decline, participants diagnosed with incident dementia over study follow-up but who did not have an interim MCR diagnosis were also excluded (n = 145). Olfactory function in the conversion from normal cognition to dementia has been previously examined.⁸ Overall, 1,119 participants were included in this analysis. Those who were excluded were older (p < 0.001) than the eligible sample but did not differ in sex (p = 0.294), years of education (p = 0.329), or number of comorbidities (p = 0.194).

The sample had a mean age of 79.14 ± 7.10 years, with 15.24 ± 3.10 years of education, and 75.1% were female. In the sample, 93.5% participants self-identified as White, 5.5% as Black, and 2.9% as Hispanic. After up to 17.2 years of follow-up time (median 3.94, interquartile range 5.15), 544 patients (48.6%) developed incident MCR (Table 1). Those who developed MCR had slightly worse semantic memory scores (p = 0.048) and more mobility impairments (p < 0.001) than those who did not develop MCR. There was no significant difference any of the other examined sample characteristics between those with an incident MCR diagnosis and those who did not develop MCR over follow-up.

Table 1.

Baseline Characteristics of the Study Population Overall and by Incident MCR Status

Scores on the BSIT at baseline ranged from 0 to 12 (mean 9.3 ± 2.0), higher indicating better. BSIT scores were related to age (r = −0.26, p < 0.001) and education (r = 0.63, p = 0.03). Females (mean score 9.5 ± 1.9) performed better than males (mean score 9.0 ± 2.2, p = 0.003). We applied a clinically relevant BSIT cut score of 8 to divide the baseline sample into those with normal olfactory function (73.4%; n = 821) and those with hyposmia (26.6%; n = 298, Table 2). Of those with hyposmia at baseline, 155 (52.0%) went on to develop MCR. Those with normal olfactory function had longer follow-up time (p < 0.001), a higher percentage of female participants (p < 0.001), a higher percentage of obese participants (p = 0.019), lower presence of parkinsonism (p = 0.006), less mobility impairment (p = 0.031), higher semantic memory scores (p < 0.001), higher global cognition scores (p < 0.001), and higher Mini-Mental Status Examination scores (p < 0.001) than those who did not. There was no difference between these groups in percent of MCR incidence, age, education, comorbidities, smoking status, depressive symptoms, or alcohol intake.

Table 2.

Baseline Characteristics of the Study Population by Olfaction Category

Olfactory Function and Incident MCR

BSIT scores at baseline were associated with an increased risk of developing MCR in adjusted models; a 1-point decrease in the BSIT score was associated with an 8% increased risk for MCR (Table 3). The association remained significant after inclusion of covariates in models 2 and 3.

Table 3.

Association of Baseline BSIT Score and Covariates With Incidence of MCR

Figure shows that participants with hyposmia (BSIT scores ≤8) at baseline had an increased risk of MCR compared with those with normal olfaction. This association remained after adjusting for covariates in models 1, 2, and 3 (Table 4).

Figure. MCR-Free Survival in Participants With Normal Olfaction and With Hyposmia at Baseline.

This Kaplan-Meier plot depicts MCR-free survival in patients with normal olfaction (n = 821; black line) and with hyposmia (n = 298; gray line) at baseline over the study period. Log-rank p value <0.001. MCR = motoric cognitive risk syndrome.

Table 4.

Association of Baseline Olfaction Category and Covariates With Incidence of MCR

To examine the effect of olfaction on MCR, independent of MCI, we conducted sensitivity analyses excluding prevalent cases of MCI (n = 221) and incident MCI cases diagnosed before incident MCR (n = 146). Of the 752 participants included, 295 (39.2%) went on to develop MCR. The baseline BSIT score was no longer associated with the risk of MCR (HR 0.95; 95% CI 0.89–1.02) after adjusting for model 1 covariates. However, the 163 (21.7%) participants with hyposmia at baseline were at an increased risk of MCR (HR 1.36; 95% CI 1.03–1.79) compared with those with normal olfaction (n = 589).

We examined whether performance on the BSIT was a very early marker of risk for MCR by excluding 222 cases with an MCR diagnosis in the first 3 years of follow-up. After adjusting for covariates in model 1, the BSIT scores were no longer associated with the risk of incident MCR (n = 322) at least 3 years after baseline (HR 0.94; 95% CI 0.88–1.00). However, participants with hyposmia at baseline (24.5%; n = 220) were at an increased risk of MCR after more than 3 years of follow-up (HR 1.37; 95% CI 1.06–1.77) compared with those with normal olfaction (n = 677).

Olfactory Function and Pathology in Patients With MCR

Study Population

Over study follow-up, there were 336 participants with an incident MCR diagnosis and an autopsy, 18 were missing pathology data, and 200 participants did not complete the BSIT at the same study visit as their first incident MCR diagnosis. Overall, 118 participants were included in this pathology analysis. Death occurred on average 3.72 (range 0.10–7.79) years after incident MCR diagnosis, and the average age at death was 92.11 ± 5.73 years. This subset of participants was 70.3% female, had an average of 15.48 ± 2.74 years of education, and a BSIT score at the time of MCR diagnosis of 8.07 ± 2.34.

Among the 118 participants included in the pathology analysis, 59 (50.0%) had normal olfactory function (BSIT scores >8) and 59 (50.0%) had hyposmia at the time of MCR diagnosis (Table 5). There were no significant differences between the groups in age at death, time to death after MCR diagnosis, sex, or years of education. Those with hyposmia and MCR had higher measures of global AD pathology (0.78 vs 0.56, p = 0.046), neurofibrillary tangles (0.81 vs 0.40, p < 0.001), density of τ tangles (9.13 vs 5.37, p < 0.001; n = 107/118), and a higher presence of Lewy body pathology compared with those with normal olfactory function and MCR. Level of neuritic plaques, diffuse plaques, β-amyloid deposition (n = 97/118), or presence of vascular pathology was not significantly different.

Table 5.

Characteristics of AD Pathology Sample by Olfaction Category at the Time of MCR Diagnosis

Global AD pathology was inversely correlated with BSIT scores at the time of MCR diagnosis (estimated coefficient −1.30, SE 0.38, p < 0.001) after controlling for age at death, sex, education, and time from testing until death (Table 6). We examined the association by type of AD pathology. Neurofibrillary tangles were robustly correlated with BSIT scores at the time of MCR diagnosis (estimated coefficient −1.23, SE 0.29, p < 0.001); neuritic plaques were also correlated, although the association was weaker (estimated coefficient −0.87, SE 0.28, p = 0.003) and diffuse plaques had no correlation with BSIT scores (p = 0.573).

Table 6.

Linear Regression Comparing Association of Global AD Pathology to BSIT Score at the Time of MCR Diagnosis

Immunohistochemical techniques were used to further investigate the relationship of plaque and tangle pathology with olfaction in patients with MCR. Density of τ tangles was inversely associated with olfactory function at the time of MCR diagnosis (estimated coefficient −0.09, SE 0.03, p = 0.002). β-amyloid deposition did not show a significant association (p = 0.661).

We examined the association between non-AD pathology factors and olfaction. The presence of Lewy body pathology was correlated with reduced olfactory function at MCR diagnosis after adjustment for covariates (estimated coefficient −1.79, SE 0.51, p < 0.001). The presence of vascular pathology was not associated with the BSIT score at MCR diagnosis (p = 0.704).

To investigate whether the associations of global AD pathology and Lewy body pathology with olfactory function were independent, we incorporated them into the same model. Both global AD pathology (estimated coefficient −1.30, SE 0.36, p < 0.001) and presence of Lewy bodies (estimated coefficient −1.79, SE 0.48, p < 0.001) remained inversely associated with the BSIT score at the time of MCR diagnosis. Similar findings were obtained when density of τ tangles and Lewy body pathology were entered in the same model (data not shown).

In sensitivity analyses to examine the association between olfactory function at MCR diagnosis and global AD pathology independent of MCI, we excluded prevalent cases of MCI (n = 21) and those diagnosed with MCI before or at the time of MCR diagnosis (n = 44). Global AD pathology was no longer associated with BSIT scores at the time of MCR diagnosis after statistical correction for multiple comparisons (p = 0.034) in this small subset of participants (n = 53). To examine a more specific measure of AD pathology, we examined the association between olfactory function at MCR diagnosis and density of τ tangles independent of MCI (estimated coefficient −0.18, SE 0.04, p < 0.001).

Discussion

Our findings indicate that the risk of incident MCR increased by 8% for each 1-point decrease in BSIT scores, and those with hyposmia had a 40% increased risk for MCR compared with those with normal olfaction. Inverse associations between olfactory dysfunction and global AD burden, neurofibrillary tangles, neuritic plaques, PHF τ tangle density, and Lewy bodies were observed. Both τ tangle density and Lewy bodies were also independently associated with olfaction in participants with MCR. To our knowledge, previous studies have not examined olfactory function in MCR as well as AD, vascular, or Lewy body pathology in patients with MCR. Our findings suggest that MCR can be considered a heterogeneous clinical risk state like MCI with subtypes that are associated with an increased risk for either Alzheimer or non-Alzheimer dementias. Poor performance on olfaction testing may be used clinically to identify an MCR subtype that is more likely to be associated with Alzheimer and Lewy body pathology.

After excluding prior MCI cases, increased risk of MCR persisted for participants with hyposmia at baseline. The correlation may not be robust enough to account for single-point value differences on the examination score given the significant overlap between MCR and MCI. However, previous analyses^27,28 and our findings show that a clinically relevant cutoff is a useful indicator of olfactory dysfunction. It is, therefore, possible that hyposmia in general may be associated with an increased dementia risk through pathways independent of MCI.

Our pathology results are consistent with those of a prior study examining pathology in older patients without cognitive impairment in this same data set.⁹ This prior study demonstrated that tangles in entorhinal cortex and CA1/subiculum regions are associated with olfactory dysfunction in older adults. Such limbic system involvement likely corresponds to the initial stages of clinical AD and associated cognitive complaints of prodromal dementia.³⁸ Neurofibrillary tangles appear in the olfactory bulb and tract in early AD³⁹ and are associated with hyposmia.⁴⁰ Our findings appear to concur within a subset of older adults diagnosed with MCR, although the regional distributions of AD pathology were not examined in this study. The association between tangle pathology and olfactory function remained after controlling for those diagnosed with MCI before MCR. Such findings support the assertion that olfactory dysfunction is correlated with tangles, irrespective of the prodromal dementia syndrome type. In our study, we found no association between β-amyloid and odor dysfunction in patients with MCR. This echoes previous findings that, after controlling for tau density, there was no association between olfactory function and postmortem β-amyloid.⁹ Studies have also demonstrated that plaques do not accumulate in the olfactory bulb and tracts until late in AD progression^12,39 and shown no correlation between olfaction and β-amyloid on PET imaging in patients with MCI.⁴¹

Gait dysfunction has been previously associated with olfactory dysfunction in cognitively normal older adults.¹⁹ Our study demonstrated that olfactory dysfunction precedes the diagnosis of this gait-based predementia syndrome. Furthermore, our findings substantiate that early olfactory dysfunction occurs in Lewy body disease,^14,15 as the presence of its pathology was inversely associated with olfactory function in patients with MCR. Given the similarities in olfactory prediction of motor symptoms, regional specificity of pathology, and our findings demonstrating independent associations of both pathologies with olfactory dysfunction, it is plausible that a common pathway connects the olfactory and gait dysfunction in Alzheimer and Lewy body diseases. In fact, a previous clinical-pathologic study observed that substantia nigra tangles were associated with gait impairment late in life regardless of dementia status or Lewy bodies.⁴² Thus, the site of pathology, rather than the specific pathology itself, may be the better indicator of olfactory and gait dysfunction.

The association between vascular dementia and olfactory dysfunction has shown mixed findings. One study has demonstrated equivalent olfactory impairment compared with AD,⁴³ whereas others have demonstrated lesser olfactory impairment.^44,45 Vascular parkinsonism can be distinguished from Parkinson disease as olfaction in vascular parkinsonism is comparable to normal controls.^46,47 It seems plausible that olfactory deficits in cerebrovascular disease may depend on the specific location, rather than the overall presence, of pathology. Our findings are consistent with this hypothesis, demonstrating no association between composite vascular pathology and olfactory function in patients with MCR. Further studies may examine categories or regional specificity of vascular pathology and its relationship with olfactory dysfunction.

This study has several limitations. Participants were predominantly White. Sex was examined as a binary variable based on self-report. Further investigations are required to examine whether these results are generalizable to other populations. The BSIT is a shortened version of the 40-item University of Pennsylvania Smell Identification Test, which may have better temporal stability and predictive validity.²⁷ Hence, BSIT may have underestimated correlations between olfaction and MCR. Cutoffs on the BSIT were not optimized based on current findings, but on initially reported cutoffs for the test.²⁷ Other studies have included anosmia (BSIT score <6) groups in analyses¹³ to examine more subtle changes in olfactory function. Unfortunately, only a small subset (n = 56) fell into the anosmia category in our sample. It is plausible that this anosmia group would have demonstrated even greater risk of incident MCR than the hyposmia grouping. Region-specific AD, Lewy body, and vascular pathologies were not examined. Finally, because this study focused on MCR, we did not examine the individual components of this predementia syndrome. But the incremental predictive validity of MCR for dementia over its individual cognitive and motoric components has been reported in MAP and other cohorts. Further research is needed to investigate how slow gait and cognitive complaints are differentially associated with hyposmia and neuropathology.

Strengths of this study include demonstration of olfactory dysfunction increasing risk of MCR even after accounting for multiple confounders in a large well-established longitudinal study with autopsy data. Exclusion of MCI cases diagnosed before or at the time of MCR diagnosis allowed us to assess AD pathology and its relation to olfactory dysfunction in patients with MCR independent of MCI. Controlling for both AD and Lewy body pathology together allowed us to ascertain both factors' independent associations with olfaction in MCR. Furthermore, because hyposmia was shown to precede MCR by more than 3 years, such olfactory dysfunction may be an early indicator of the transition from normal cognitive function to MCR.

The present findings demonstrate that difficulty in identifying odors precedes the transition from normal cognition to MCR. The olfactory deficits may be mediated by either neurofibrillary tangles or Lewy body pathology. This transition is important as patients with MCR are already at an increased risk for dementia, disability, falls, and death.⁴⁸ As treatments are developed to prevent or delay the onset of dementia, it may be necessary to intervene before MCR is diagnosed. The results of this and other prospective studies^5,21 indicate that it may be plausible to identify cognitively normal older individuals who are at an increased risk for transitioning first to MCR and then to dementia as potential candidates for such interventions.

Glossary

AD

Alzheimer disease

BMI

body mass index

BSIT

Brief Smell Identification Test

HR

hazard ratio

MAP

Memory and Aging Project

MCI

mild cognitive impairment

MCR

motoric cognitive risk syndrome

PHF

paired helical filament

Appendix. Authors

Study Funding

The research was supported by the National Institute On Aging (NIA) under award number 1R01AG057548-01A1. Funding agencies for the participating cohorts are as follows: The Memory & Aging Project is supported by the NIH/National Institute on Aging grants (R01AG17917) and the Illinois Department of Public Health.

Disclosure

N.L. Kravatz reports no relevant disclosures. E. Ayers received funding support from the National Institute on Aging under award number 1R01AG057548-01A1. D.A. Bennett received funding support from the NIH/National Institute on Aging grants (R01AG17917) and the Illinois Department of Public Health. J. Verghese received funding support from the National Institute on Aging under award number 1R01AG057548-01A1. Go to Neurology.org/N for full disclosures.