Abstract
Introduction:
We investigated associations of impaired odor identification with signs of early Alzheimer’s disease (AD) and Lewy Body diseases (LBD).
Methods:
In 787 community-sampled adults aged ≥65 without dementia or Parkinson’s disease, we evaluated cross-sectional associations of olfaction (Brief Smell Identification Test, BSIT) with five cognitive domain scores, gait, tone, tremor, REM sleep behavior disorder (RBD) symptoms, and plasma biomarkers of neurodegeneration
Results:
In adjusted regression models, worse BSIT scores were associated with lower cognitive scores in all domains, slowed gait, resting tremor, and RBD-like symptoms. The association between BSIT and memory was stronger in APOE4 carriers. In cognitively unimpaired participants (Clinical Dementia Rating=0), most associations remained significant. BSIT scores were associated with plasma p-tau217 and NfL in exploratory analyses.
Conclusion:
Odor identification is associated with cognition in an undifferentiated pattern and with Parkinsonian signs in older adults, even among those with normal cognition. The stronger association in APOE4 carriers potentially suggests an emerging AD-like pathway, consistent with exploratory plasma biomarker analyses. In sum, evidence points to olfactory dysfunction as relevant for both AD and LBD-associated future risk.
Keywords: Olfaction, cognitive domains, aging, Lewy Body Disorders, Alzheimer’s disease
Introduction
Impairment in olfaction is a known marker of aging and neurodegenerative diseases, particularly Alzheimer’s disease (AD) and Lewy Body diseases (LBD), which include Parkinson’s disease (PD), Parkinson’s disease dementia (PDD), and dementia with Lewy bodies (DLB)1. Both AD and LBD are characterized by long prodromal periods, which can include olfaction impairment that is present years before other clinical features of disease manifest. Olfaction impairment has been reported to predict transition from mild cognitive impairment (MCI) to dementia2,3, future cognitive decline in unimpaired older adults2–4, and mortality5.
Impaired olfaction has also been associated with poor performance on a wide variety of motor tasks in healthy older adults6,7 including slowed gait cross-sectionally and physical function decline over 8 years8. Olfactory dysfunction is also associated with rapid eye movement (REM) sleep behavior disorder (RBD), an early marker of α-synucleinopathy9. A meta-analysis found associations supporting impaired olfaction as a durable biomarker for RBD and prodromal PD10. In individuals with RBD, hyposmia was the strongest clinical predictor of abnormal CSF aggregation of synuclein11, and a combination of probable RBD and olfactory deficits was highly specific for predicting presence of post-mortem Lewy type α-synucleinopathy12.
Many of the above findings regarding olfactory impairment resulted from clinic-based or volunteer sample studies, although there have been some investigations in population-based cohorts as well3,4. Population studies are informative as they test external validity of clinic findings and are somewhat less vulnerable to selection biases than clinic and volunteer samples. They can be informative about the distributions of commonly utilized clinic measures by key demographic factors in populations of interest. Regarding cognition, relatively few studies probing olfaction included multiple domains, which may shed light into specific neurodegenerative risk pathways, e.g., memory and language potentially reflecting AD risk; visual-spatial abilities, attention, processing speed, executive functions potentially reflecting LBD risk13. In addition, the APOE4 allele, which increases risk for neurodegenerative diseases and dementia, is not well characterized regarding its potential role as an effect modifier of clinical and cognitive associations with olfactory deficits14.
In the present study we leveraged data from two community-based studies to examine cross-sectional relationships between olfactory identification performance and multi-domain cognition, motor function, RBD-like symptoms, and APOE genotype in a cohort of older adults without dementia or Parkinson’s disease. The goals were to evaluate 1) whether olfaction associations could be detected in this population cohort; 2) if olfaction-cognition association patterns were more consistent with an AD or a LBD process; 3) if the APOE4 allele was a significant effect modifier; and 4) whether olfaction associations observed in the larger cohort, including MCI, were also seen in a cognitively normal subset, reflecting very earliest of risk. Finally, in a post-hoc exploratory analysis in a subset of participants, we examined associations between olfaction and plasma neurodegeneration-related biomarkers preceding assessment, with time intervals between 0 and ~10 years.
Methods
Study Setting and Participants
The Monongahela-Youghiogheny Heathy Aging Team (MYHAT) and 15104 Seniors Project (SP15104) are community-based prospective studies which identify MCI and its risk and protective factors in communities of relatively low socioeconomic status in southwestern Pennsylvania. Recruitment for the population-based MYHAT study by age-stratified random sampling of individuals ≥65 years from publicly available voter registration lists occurred in 2006–2008 and 2016–2019 with the later cohort aged 65–74. Recruitment for the community volunteer-based SP15104 took place in 2019–2025 using intensive community engagement in three towns with ~60% African American representation, within the larger MYHAT area. Details have been previously published15. Eligibility criteria at study entry for all cohorts included being the target age, residence in the designated towns, not residing in long-term care, sufficient vision and hearing to undergo neuropsychological testing, decisional capacity, and scoring ≥21/30 on the age-education-corrected Mini-Mental State Examination (MMSE)16. A total of 2916 participants completed the full baseline assessment and were invited to undergo annual re-assessment.
Study Assessments
Demographics, health history, and lifestyle behaviors.
Participants self-reported their health history, including but not limited to Parkinson’s disease, recent COVID-19 infection, and tobacco use.
Olfactory assessment.
In 2021, we added the Brief Smell Identification Test (BSIT)17 to the study assessment. The BSIT has been utilized widely across demographic and ethnic communities18,19. Familiar odors microencapsulated on a card are revealed when scratched with a pencil. Participants sniff the odor and identify it from a choice of 4 responses. One point is given for each correct response, lower scores indicating worse odor identification.
Clinical Dementia Rating.
Following each annual assessment, participants were evaluated by trained interviewers using the Clinical Dementia Rating® (CDR)20, a functional assessment of cognition. We categorized a score of 0 as cognitively unimpaired; 0.5 as mild cognitive impairment (MCI); ≥1 as dementia. Individuals with CDR≥1 were excluded from the present analyses.
Apolipoprotein E (APOE) genotyping:
Participants provided optional blood or saliva samples for genomic testing, including APOE genotyping.
Neuropsychological testing.
Participants completed a battery of neuropsychological tests tapping 5 cognitive domains: attention and processing speed; memory; language; executive functions; and visuospatial abilities. Composite scores for each domain were calculated by first standardizing each test according to the baseline mean and standard deviation for each recruitment cohort and then averaging the standardized test scores within each domain. Higher composite scores indicate better performance.
General cognition.
The MMSE was used to assess overall cognitive status, with a score <24 out of 30 categorized as impairment.
Gait.
Gait speed was assessed with the Timed Up and Go (TUG) test. We used a cutoff of >13 seconds to complete the TUG to represent slowed gait21. Gait sequence and posture were also assessed, and for the present study, we focused on shuffling and festination, two features of Parkinsonian gait.
Motor tone.
The presence of cogwheel rigidity, a ratchet-like jerkiness throughout the range of motion, was assessed in the right and left wrists and elbows.
Resting tremor.
With the participant at rest, tremors of the head and extremities were evaluated. For the current study, we considered tremors of either extremities or head separately and together.
REM sleep behavior disorder (RBD)-like symptoms:
Participants were asked if they had ever been told, or suspected themselves, that they seem to “act out their dreams” while asleep. If yes, they were asked if this had happened in the preceding year. This single question has been validated as a screening tool for REM sleep behavior disorder22.
Plasma biomarkers of AD and related disorders (ADRD).
A subset of participants provided a venous blood sample at various timepoints between 2014 and 2021. Using Single Molecule Array (SIMOA) technology, plasma specimens were tested for p-tau181 using the p-tau181 V2 Advantage (#103714) assay, and for NfL, GFAP, p-tau217, Aβ42, and Aβ40 with the Neurology 4-Plex E (#103670) commercial assays, as detailed previously.23 Due to participant refusal and the exclusion of values below assay detection limits, 432 participants had plasma biomarker values (range n=370–422 per biomarker).
Statistical Methods
Sample Description.
Comparisons between BSIT groups used two-sample t-test and Wilcoxon rank-sum test for continuous variables with normal or non-normal distributions, respectively. Pearson’s chi-squared test was used for categorical variables, and Fisher’s exact test when any cell counts were <5.
Modeling.
We examined BSIT scores both as a continuous variable and dichotomized at a score of <9 out of 12 to categorize abnormal odor identification24. We used linear regression models to assess associations between BSIT and cognitive composite scores. We fit logistic regression models for binary outcomes. We reported estimates as beta coefficients (for continuous outcomes) or odds ratios (ORs, for binary outcomes) with corresponding 95% confidence intervals (CIs).
For all associations tested, we fit univariable models followed by multivariable models adjusting for age, sex, education level, and smoking status (never vs. former vs. current; see Results). For domain-specific cognitive composite scores, we additionally examined the main effect of APOE4 carrier status and its potential effect modifications of those associations by including APOE4*BSIT interaction terms in the models, tested by likelihood ratio tests.
To examine cross-sectional associations between olfactory function and plasma biomarker concentrations in a post-hoc analysis within a subset of participants, we employed linear regression models with biomarker values as outcomes and continuous BSIT scores as the predictor, adjusting for age, sex, race, education, and APOE4 carrier status. All biomarker values (Aβ40, Aβ42, Aβ42/40 ratio, p-tau181, p-tau217, NfL, and GFAP) were natural log-transformed to address right-skewed distributions and improve model fit. See Supplemental Methods for details and sensitivity analyses.
All analyses were conducted using the entire sample (CDR<1) and repeated among cognitively unimpaired participants (CDR=0) to assess the impact of cognitive-functional status. All statistical analyses were performed using R Statistical Software version 4.5.1 (The R Foundation for Statistical Computing, Vienna, Austria) with significance defined as a two-tailed p-value <0.05. As analyses were exploratory, we did not apply multiple comparison correction methods25,26.
Results
Participant Characteristics
We report here on the initial BSIT data from 787 participants, excluding 5 participants with dementia (CDR≥1), 3 with missing CDR, and 10 who endorsed diagnoses of PD. Of these, 670 (85.1%) had CDR=0 (cognitively unimpaired) and 117 (14.9%) had CDR=0.5 (MCI).
Among all participants, 241 (30.6%) demonstrated abnormal BSIT (BSIT<9). Compared to participants with normal BSIT, those with abnormal scores were significantly older, had a higher proportion of males, were less educated, had higher rates of smoking, and had a higher proportion of MCI (CDR=0.5) (Table 1).
Table 1.
Participant characteristics by BSIT groups
| Normal BSIT: Score ≥9 (N=546) | Abnormal BSIT: Score <9 (N=241) | Total (N=787) | p value | |
|---|---|---|---|---|
| Age, years | 74.0 (71.0, 78.0) | 78.0 (73.0, 84.0) | 75.0 (71.0, 80.0) | < 0.001 1 |
| Female | 370 (67.8%) | 137 (56.8%) | 507 (64.4%) | 0.003 2 |
| Race non-white | 110 (20.1%) | 58 (24.1%) | 168 (21.3%) | 0.2162 |
| Education >HS | 354 (64.8%) | 138 (57.3%) | 492 (62.5%) | 0.043 2 |
| APOE4 carrier | 107 (24.6%) | 40 (20.3%) | 147 (23.3%) | 0.2372 |
| Missing | 111 | 44 | 155 | |
| Smoking status | 0.024 2 | |||
| Never | 264 (48.4%) | 98 (40.7%) | 362 (46.0%) | |
| Former | 239 (43.8%) | 111 (46.1%) | 350 (44.5%) | |
| Current | 43 (7.9%) | 32 (13.3%) | 75 (9.5%) | |
| Covid-19 positive | 102 (18.8%) | 50 (20.8%) | 152 (19.4%) | 0.4962 |
| Missing | 2 | 1 | 3 | |
| MCI | 55 (10.1%) | 62 (25.7%) | 117 (14.9%) | < 0.001 2 |
| Attention | 0.09 (0.76) | −0.18 (0.75) | 0.01 (0.77) | < 0.001 3 |
| Missing | 1 | 1 | 2 | |
| Executive Function | 0.11 (0.70) | −0.30 (0.84) | −0.02 (0.77) | < 0.001 3 |
| Missing | 1 | 0 | 1 | |
| Language | 0.11 (0.70) | −0.26 (0.84) | −0.00 (0.77) | < 0.001 3 |
| Missing | 1 | 2 | 3 | |
| Memory | 0.46 (0.89) | −0.02 (1.09) | 0.31 (0.98) | < 0.001 3 |
| Missing | 2 | 3 | 5 | |
| Visuospatial | 0.11 (0.81) | −0.20 (0.93) | 0.02 (0.86) | < 0.001 3 |
| Missing | 31 | 40 | 71 | |
| MMSE | 28.0 (26.0, 29.0) | 27.0 (25.0, 29.0) | 28.0 (26.0, 29.0) | < 0.001 1 |
| Cognitive impairment | 29 (5.3%) | 32 (13.3%) | 61 (7.8%) | < 0.001 2 |
| Timed Up and Go | 13.00 (11.00, 15.25) | 14.00 (12.00, 18.00) | 13.00 (12.00, 16.00) | < 0.001 1 |
| Missing | 94 | 75 | 169 | |
| Slowed gait | 183 (40.5%) | 96 (57.8%) | 279 (45.1%) | < 0.001 2 |
| Missing | 94 | 75 | 169 | |
| Shuffling gait | 14 (2.8%) | 12 (5.7%) | 26 (3.7%) | 0.0552 |
| Missing | 43 | 32 | 75 | |
| Festinating gait | 5 (1.0%) | 0 (0.0%) | 5 (0.7%) | 0.3294 |
| Missing | 43 | 32 | 75 | |
| Left-side cogwheel rigidity | 5 (1.7%) | 4 (2.7%) | 9 (2.0%) | 0.4904 |
| Missing | 249 | 92 | 341 | |
| Right-side cogwheel rigidity | 9 (3.0%) | 6 (4.0%) | 15 (3.4%) | 0.5904 |
| Missing | 250 | 90 | 340 | |
| Either-side cogwheel rigidity | 10 (3.4%) | 6 (4.0%) | 16 (3.6%) | 0.7894 |
| Missing | 251 | 92 | 343 | |
| Resting tremor in head | 5 (0.9%) | 2 (0.8%) | 7 (0.9%) | 1.0004 |
| Missing | 10 | 2 | 12 | |
| Resting tremor in extremities | 31 (5.8%) | 24 (10.0%) | 55 (7.1%) | 0.034 2 |
| Missing | 11 | 2 | 13 | |
| Resting tremor in extremities or head | 33 (6.2%) | 26 (10.9%) | 59 (7.6%) | 0.023 2 |
| Missing | 11 | 2 | 13 | |
| Ever having RBD-like symptoms | 38 (7.1%) | 29 (12.1%) | 67 (8.6%) | 0.021 2 |
| Missing | 10 | 2 | 12 | |
| RBD-like symptoms in last year | 16 (3.0%) | 19 (7.9%) | 35 (4.5%) | 0.002 2 |
| Missing | 11 | 2 | 13 |
Continuous variables (age, MMSE score, and timed up and go) are presented as median (Q1, Q3) and compared using Wilcoxon rank-sum tests; continuous cognitive composite scores are presented as mean (SD) and compared by the two-sample t-test. Categorical variables are reported as frequency (%) and compared using Pearson’s Chi-squared test if all expected cell counts are greater than 5, or Fisher’s exact test otherwise. All p-values from tests are provided, with statistical significance defined as a two-sided p-value <0.05.
Cutoffs: Abnormal BSIT defined as BSIT score <9; MCI defined as Clinical Dementia Rating (CDR)=0.5; Cognitive impairment defined as MMSE score <24; Slowed gait defined as timed up and go >13 seconds.
Abbreviations: BSIT, Brief Smell Identification Test; APOE4, Apolipoprotein E4 allele; MCI, mild cognitive impairment; MMSE, Mini-Mental State Examination; RBD, REM sleep behavior disorder.
Superscripts:
Wilcoxon rank-sum test
Pearson’s Chi-squared test
Two-sample t-test
Fisher’s exact test
Unadjusted Associations
In unadjusted models with continuous BSIT scores as outcomes, we found significant positive associations with all cognitive domain scores. Note that the likelihood maximization algorithm failed to converge for modeling festinating gait due to complete or quasi-complete separations. Better BSIT scores were associated with higher scores of attention, executive functions, language, memory, and visuospatial abilities, and lower odds of cognitive impairment (MMSE<24). Worse BSIT scores were also associated with higher odds of slowed gait (TUG>13s), shuffling gait, resting tremor in extremities, resting tremor in extremities or head, ever having RBD-like symptoms, and RBD-like symptoms in the past year.
Unadjusted model results with binarized BSIT scores as outcomes yielded consistent findings, except that the positive association with shuffling gait was not statistically significant.
Adjusted Associations
Figure 1 presents findings adjusted for age, sex, race, education and smoking status. Of note, the model convergence problems also occurred for shuffling gait, festinating gait, and cogwheel rigidity Most adjusted associations between BSIT performance and outcomes of interest remained statistically significant, though effect sizes were generally attenuated. For both continuous and binary BSIT scores, associations with all cognitive domains were significant. For other outcomes, higher BSIT scores were associated with lower odds of cognitive impairment, slowed gait, resting tremor in extremities or head, and having RBD-like symptoms both ever and in the past year. Additionally, participants with higher continuous BSIT scores had lower odds of resting tremor in extremities.
Figure 1.
Forest plots showing adjusted associations between cognitive and clinical outcomes and BSIT scores. Models were adjusted for continuous age, sex, race, education groups, and smoking status. BSIT (continuous and binary) was the main exposure. Results are presented as β-coefficient estimates with 95% confidence intervals (CI) for continuous outcomes (left panel), and odds ratios (OR) with 95% CI for binary outcomes (right panel). Circles represent continuous BSIT associations, while squares represent abnormal BSIT associations. Filled symbols indicate statistically significant associations (p<0.05). Values shown on the right of each panel display the estimate and 95% CI for each association. Reference lines are shown at β=0 for continuous outcomes and OR=1 for binary outcomes. Cutoffs: Abnormal BSIT defined as BSIT scores <9; Cognitive impairment defined as MMSE score <24; Slowed gait defined as timed up and go >13 seconds. Abbreviations: BSIT, Brief Smell Identification Test; RBD, REM sleep behavior disorder.
Effect Modification by APOE4 Carriership
We found evidence of significant effect modification by APOE4 carriership (Supplemental Table 1). For the memory domain, an interaction was observed indicating APOE4 carriership strengthened the association between abnormal BSIT and lower memory scores. A significant interaction effect was also found with cognitive impairment (MMSE<24), indicating that the higher odds of cognitive impairment in the abnormal BSIT group, compared to the normal group, was stronger among APOE4 carriers. However, these findings did not reach statistical significance in the models with continuous BSIT scores (p values >0.05).
Post-Hoc Analyses: Plasma Biomarkers
Following the main results, we investigated plasma ADRD biomarkers and their association with BSIT performance, considered exploratory because of methodological limitations (see Supplemental Methods and Discussion). Compared to the whole analytic sample, participants with available biomarker data were older (median 75.0 vs. 73.0 years), more likely to report white race (95.4% vs. 58.3%), and more educated (66.7% vs. 57.5% >HS). APOE4 carriership, MCI proportion, BSIT score, and proportion of abnormal BSIT were comparable.
Plasma biomarker assessments preceded BSIT testing, with time intervals ranging from 0 to approximately 10 years (Q1=3.20 years, Q3=5.92 years). In a series of sensitivity analyses restricting observations to varying temporal gaps (Supplemental Figure 1) and in models adjusting for date difference between BSIT and blood draw as a main effect (Supplemental Table 2), BSIT scores were significantly inversely associated with plasma p-tau217and NfL. No significant associations were observed between BSIT scores and Aβ40, Aβ42, Aβ42/40 ratio, GFAP, or p-tau181. See Supplemental Results for further details.
Sensitivity Analyses among Cognitively Unimpaired Participants (CDR=0)
Among cognitively unimpaired participants (n= 670), 179 (26.7%) had abnormal BSIT scores (BSIT<9). They were significantly older, had a higher proportion of males, and had higher rates of smoking (Supplemental Table 3) compared to participants with normal BSIT. In unadjusted models, almost all the associations observed in the whole sample were still significant, with generally smaller effect sizes compared to the full sample. Exceptions included shuffling gait, resting tremor in extremities and in extremities or head, which were no longer significant.
Figure 2 presents estimated results adjusted for age, sex, race, education, and smoking status among cognitively unimpaired participants. Better BSIT performance was consistently associated with better cognitive scores for all domains and lower odds of cognitive impairment. It was also consistently associated with lower odds of slowed gait, ever having RBD-like symptoms, and past year RBD-like symptoms, and for the continuous BSIT score only, with lower odds of resting tremor in extremities or head.
Figure 2.
Forest plots showing adjusted associations between cognitive and clinical outcomes and BSIT scores among cognitively unimpaired participants. Models were adjusted for continuous age, sex, race, education groups, and smoking status. BSIT (continuous and binary) was the main exposure. Results are presented as β-coefficient estimates with 95% confidence intervals (CI) for continuous outcomes (left panel), and odds ratios (OR) with 95% CI for binary outcomes (right panel). Circles represent continuous BSIT associations, while squares represent abnormal BSIT associations. Filled symbols indicate statistically significant associations (p<0.05). Values shown on the right of each panel display the estimate and 95% CI for each association. Reference lines are shown at β=0 for continuous outcomes and OR=1 for binary outcomes. Cutoffs: Abnormal BSIT defined as BSIT scores <9; Cognitive impairment defined as MMSE score <24; Slowed gait defined as timed up and go >13 seconds. Abbreviations: BSIT, Brief Smell Identification Test; RBD, REM sleep behavior disorder.
APOE4 no longer modified any associations between memory or MMSE and abnormal BSIT.
In the ADRD plasma biomarker analysis within the cognitively unimpaired subset, controlling for demographics and temporal gap between blood collection and BSIT, BSIT scores showed significant inverse associations with multiple biomarkers (Supplemental Table 2 and Supplemental Figure 2). Better BSIT scores were associated with lower concentrations of Aβ42, p-tau217, and NfL, corresponding to approximately 1.8%, 3.2%, and 3.0% lower concentrations of Aβ42, p-tau217, and NfL, respectively. The association with Aβ42 in cognitively unimpaired participants contrasts with the null findings in the full sample. The robust associations with p-tau217 and NfL were slightly stronger than those observed in the full sample.
Discussion
In this community-based cohort of dementia-free older adults from a small-town U.S. area, we report associations between poorer odor identification and worse attention, memory, language, executive functions, and visuospatial abilities, impairment by MMSE, presence of resting tremor in extremities or head, self-reported REM sleep behavioral symptoms, and slowed gait. Interestingly, while APOE4 carriership was not directly associated with impaired olfaction, the interaction between APOE4 and odor identification was significant on memory domain scores and MMSE<24, in other words, the association between poorer olfaction and worse memory and MMSE was stronger in APOE4 carriers than in non-carriers.
Olfactory dysfunction is associated with pathologic signs of brain aging14, including smaller hippocampal volume and cortical thickness, even in those with normal cognition27. Neurofibrillary tangles in the olfactory bulb, from which neurons project to multiple brain structures damaged in AD, correlate with odor identification deficits28, and impaired olfaction in LBD may result from α-synuclein pathology in the olfactory bulb1.
An aim was to evaluate whether patterns of associations with olfaction may more closely resemble correlates of an emerging AD vs. α-synuclein process, among different cognitive domains at the group level. Results indicate an undifferentiated pattern with odor identification directly associated with performance on all cognitive domains, and with clinical measures of MCI. In previous cohort studies examining multiple cognitive domains, findings were somewhat consistent. In the Rush Memory and Aging Project, participants without dementia and with olfactory impairment had a faster decline in episodic memory, visuospatial ability, perceptual speed, and semantic memory, but not working memory, than those with normal olfaction14. The Mayo Clinic Study of Aging found both cross-sectional and longitudinal associations between baseline olfaction and memory, executive function and language domain scores, but not visuospatial abilities. Poorer olfaction predicted incident amnestic MCI but not incident non-amnestic MCI, suggestive of a stronger underlying AD trajectory3. Notably, in the present cross-sectional findings, every measure of cognition was associated with olfaction, also within the cognitively unimpaired subgroup.
The effect modifications by APOE4 carrier status, however, were specific to memory scores, with ε4 carriers showing significantly stronger olfaction-cognition associations than non-carriers. Memory impairment is the hallmark early harbinger of AD, and APOE4 increases AD lifetime risk and promotes earlier onset and greater levels of amyloid-beta accumulation29. However, memory is also frequently impaired in DLB, although usually to a milder degree, and associated with concomitant AD pathology30. The APOE4 allele is also associated with cognitive decline in DLB31 and PD32, and evidence supports a direct potentiating effect of APOE4 on α-synuclein spreading and risk for cognitive impairment, independent of effects on amyloid-beta or tau33. Thus, in the present results, the APOE4 augmentation of the olfaction-memory association suggests a neurodegenerative pathway in which poor olfaction may signal early prodromal AD. However, results do not clearly differentiate between LBD and AD early pathways. Regardless, APOE4 status appears to be an important factor that potentiates risk for neurodegenerative disease from olfactory identification deficits, consistent with some other studies. For example, a recent population-based study in Sweden reported associations between deficits in odor identification and neurodegenerative mortality after 6 years which was significantly increased with APOE4 carriage5.
Regarding olfactory identification associations with gait speed and self-reported REM sleep behavioral symptoms, these findings are consistent with existing literature. In the Health ABC study, poorer olfaction was associated with slower walking speeds in well-functioning older adults and predicted faster decline in walking speed over follow-up8. Olfactory dysfunction was also associated with a wide range of physical performance tests in the Baltimore Longitudinal Study of Aging such as balance, fine motor function, mobility, and manual dexterity7. In patients with isolated diagnosed RBD, olfactory dysfunction predicted transition to LBD over five years with high accuracy34. Another recent clinical study showed that almost all patients with both isolated RBD and olfactory dysfunction had a positive CSF alpha-synuclein seed amplification assay, suggesting the combination of these two predictors are likely especially effective for screening for prodromal LBD11. These associations have been well documented in both clinic-based and population studies, but rarely in population studies with in-depth assessment and characterization of cognitively normal status, as the present results show.
Regarding olfaction associations with observed resting tremor on neurologic exam, a significant clinical literature supports olfactory dysfunction as an approach to differentiate Parkinson’s disease from essential and other types of tremor35. Prevalence of LBD and other movement disorders has been assessed and reported in population studies36,37. However, we are unaware of population studies like MYHAT with in-depth normal cognition characterization and neurologic exam findings, reporting associations between olfaction and resting tremor of the head and extremities in cognitively unimpaired older adults. Of note, we could not evaluate other Parkinsonian signs of festinating gait, shuffling gait, and cogwheel rigidity, as these signs were too infrequently observed.
We conducted a post-hoc analysis of plasma ADRD biomarkers in association with odor identification, considered highly exploratory because: 1) the number of available samples was in a smaller subset, and 2) the timing of blood drawn relative to BSIT was distributed out to a maximum of 10 years prior to assessment, limiting inferences about temporal associations. We observed stability in the associations over time, however, providing confidence in the results. Consistent with recent studies27,38, we found associations between poorer olfaction and higher plasma p-Tau217 and NfL in the whole sample and among cognitively unimpaired participants, as well as with Aβ42 in the cognitively unimpaired subset. Elevated plasma p-Tau217 is strongly associated with presence of Aβ plaques and neurofibrillary tangles39 and is thus further evidence for an AD-dominant pathway, while elevated plasma NfL is a non-specific marker of neurodegeneration40. We interpret these results with caution.
In sum, the present findings support prior literature indicating poorer odor identification is an early indicator of neurodegenerative risk, including risk for synucleinopathies, for some older adults. They extend prior findings by 1) demonstrating olfaction associations with cognition, physical function and Parkinson signs and symptoms in this specific population cohort in a small-town, economically disadvantaged U.S. area, a population under-represented in most academic research; and 2) observing nearly all associations remained significant in a cognitively unimpaired subset, suggesting evidence of pre-clinical disease pathways detectable in a ‘real-world’ setting.
Limitations include the cross-sectional design and as expected in a population sample, a relatively small MCI group. Only self-reported RBD-like symptoms were available, so this is a proxy for diagnosed RBD but likely resulted in missed true cases and false positives. Finally, ADRD biomarker data collection was limited as described above. Future work will include longitudinal follow-up for cognitive, motor, neurological and health outcomes, as well as ADRD blood biomarker sampling subsequent to olfaction assessment, to further probe the role of olfaction as a very early preclinical predictor of neurodegeneration and its potential for population screening.
Supplementary Material
Supplemental Digital Content.docx
Acknowledgements
The authors thank the participants of MYHAT and the 15104 Seniors Project for their time and contributions and all study personnel for their efforts in project coordination, data collection, and processing.
Conflicts of interest and sources of funding:
The work presented here was funded by R37AG023651 and R01AG05854901 from the National Institute on Aging, National Institutes of Health, US DHHS. Over the last two years, TKK has consulted for/served on advisory boards for Quanterix Corporation, SpearBio Inc., Neurogen Biomarking LLC, Alzheon, and Siemens Healthineers, and has received honorarium from Cell Signaling Technology. Outside the submitted work, TKK has received royalties from Bioventix for the transfer of specific antibodies and assays to third party organizations. TKK is an inventor on patents and provisional patents regarding biofluid biomarker methods, targets and reagents/compositions, that may generate income for the institution and/or self should they be licensed and/or transferred to another organization. For the remaining authors, no conflicts were declared.
References
- 1.Beach TG, Adler CH, Zhang N, et al. Severe hyposmia distinguishes neuropathologically confirmed dementia with Lewy bodies from Alzheimer’s disease dementia. PLoS One. 2020;15(4):e0231720. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Knight JE, Yoneda T, Lewis NA, et al. Transitions Between Mild Cognitive Impairment, Dementia, and Mortality: The Importance of Olfaction. J Gerontol A Biol Sci Med Sci. Jul 8 2023;78(7):1284–1291. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Roberts RO, Christianson TJ, Kremers WK, et al. Association Between Olfactory Dysfunction and Amnestic Mild Cognitive Impairment and Alzheimer Disease Dementia. JAMA Neurol. Jan 2016;73(1):93–101. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Devanand DP, Lee S, Manly J, et al. Olfactory deficits predict cognitive decline and Alzheimer dementia in an urban community. Neurology. Jan 13 2015;84(2):182–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Ruane R, Lampert O, Larsson M, et al. Olfactory Deficits and Mortality in Older Adults. JAMA Otolaryngol Head Neck Surg. Jun 1 2025;151(6):558–566. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Yuan Y, Li C, Luo Z, et al. Olfaction and Physical Functioning in Older Adults: A Longitudinal Study. J Gerontol A Biol Sci Med Sci. Aug 12 2022;77(8):1612–1619. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Tian Q, Resnick SM, Studenski SA. Olfaction Is Related to Motor Function in Older Adults. J Gerontol A Biol Sci Med Sci. Aug 1 2017;72(8):1067–1071. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Yuan Y, Chamberlin KW, Li C, et al. Olfaction and Mobility in Older Adults. JAMA Otolaryngol Head Neck Surg. Mar 1 2024;150(3):201–208. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Chan PC, Lee HH, Hong CT, et al. REM Sleep Behavior Disorder (RBD) in Dementia with Lewy Bodies (DLB). Behav Neurol. 2018;2018:9421098. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Lyu Z, Zheng S, Zhang X, et al. Olfactory impairment as an early marker of Parkinson’s disease in REM sleep behaviour disorder: a systematic review and meta-analysis. J Neurol Neurosurg Psychiatry. Mar 2021;92(3):271–281. [DOI] [PubMed] [Google Scholar]
- 11.Siderowf A, Concha-Marambio L, Lafontant DE, et al. Assessment of heterogeneity among participants in the Parkinson’s Progression Markers Initiative cohort using alpha-synuclein seed amplification: a cross-sectional study. Lancet Neurol. May 2023;22(5):407–417. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Tremblay C, Adler CH, Shill HA, et al. Predicting Post-Mortem alpha-Synuclein Pathology by the Combined Presence of Probable REM sleep behavior disorder and Hyposmia. Mov Disord Clin Pract. Feb 2025;12(2):157–165. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Hemminghyth MS, Chwiszczuk LJ, Rongve A, et al. The Cognitive Profile of Mild Cognitive Impairment Due to Dementia With Lewy Bodies-An Updated Review. Front Aging Neurosci. 2020;12:597579. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Dintica CS, Marseglia A, Rizzuto D, et al. Impaired olfaction is associated with cognitive decline and neurodegeneration in the brain. Neurology. Feb 12 2019;92(7):e700–e709. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Ganguli M, Jacobsen E, Song R, et al. Social determinants of health and mild cognitive impairment in a diverse community sample. J Am Geriatr Soc. Feb 2025;73(2):367–378. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Mungas D, Marshall SC, Weldon M, et al. Age and education correction of Mini-Mental State Examination for English and Spanish-speaking elderly. Neurology. Mar 1996;46(3):700–6. [DOI] [PubMed] [Google Scholar]
- 17.Doty RL, Marcus A, Lee WW. Development of the 12-item Cross-Cultural Smell Identification Test (CC-SIT). Laryngoscope. Mar 1996;106(3 Pt 1):353–6. [DOI] [PubMed] [Google Scholar]
- 18.Hawkins KA, Pearlson GD. Age and gender but not common chronic illnesses predict odor identification in older African Americans. Am J Geriatr Psychiatry. Sep 2011;19(9):777–82. [DOI] [PubMed] [Google Scholar]
- 19.Menon C, Westervelt HJ, Jahn DR, et al. Normative performance on the Brief Smell Identification Test (BSIT) in a multi-ethnic bilingual cohort: a Project FRONTIER study. Clin Neuropsychol. 2013;27(6):946–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Morris JC. The Clinical Dementia Rating (CDR): current version and scoring rules. Neurology. Nov 1993;43(11):2412–4. [DOI] [PubMed] [Google Scholar]
- 21.Shumway-Cook A, Brauer S, Woollacott M. Predicting the probability for falls in community-dwelling older adults using the Timed Up & Go Test. Phys Ther. Sep 2000;80(9):896–903. [PubMed] [Google Scholar]
- 22.Postuma RB, Arnulf I, Hogl B, et al. A single-question screen for rapid eye movement sleep behavior disorder: a multicenter validation study. Mov Disord. Jun 2012;27(7):913–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Ferreira PCL, Zhang Y, Snitz B, et al. Plasma biomarkers identify older adults at risk of Alzheimer’s disease and related dementias in a real-world population-based cohort. Alzheimers Dement. Oct 2023;19(10):4507–4519. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Doty RL. The Brief Smell Identification Test Administration Manual. Sensonics, Inc.; 2001. [Google Scholar]
- 25.R Core Team. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing; 2024. https://www.R-project.org/ [Google Scholar]
- 26.Garcia-Perez MA. Use and misuse of corrections for multiple testing. Methods in Psychology. 2023;8:100120. [Google Scholar]
- 27.Dong Y, Li Y, Liu K, et al. Anosmia, mild cognitive impairment, and biomarkers of brain aging in older adults. Alzheimers Dement. Feb 2023;19(2):589–601. [DOI] [PubMed] [Google Scholar]
- 28.Wilson RS, Arnold SE, Schneider JA, et al. The relationship between cerebral Alzheimer’s disease pathology and odour identification in old age. J Neurol Neurosurg Psychiatry. Jan 2007;78(1):30–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Mishra S, Blazey TM, Holtzman DM, et al. Longitudinal brain imaging in preclinical Alzheimer disease: impact of APOE epsilon4 genotype. Brain. Jun 1 2018;141(6):1828–1839. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Nedelska Z, Schwarz CG, Lesnick TG, et al. Association of Longitudinal beta-Amyloid Accumulation Determined by Positron Emission Tomography With Clinical and Cognitive Decline in Adults With Probable Lewy Body Dementia. JAMA Netw Open. Dec 2 2019;2(12):e1916439. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Guerreiro R, Ross OA, Kun-Rodrigues C, et al. Investigating the genetic architecture of dementia with Lewy bodies: a two-stage genome-wide association study. Lancet Neurol. Jan 2018;17(1):64–74. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Liu G, Peng J, Liao Z, et al. Genome-wide survival study identifies a novel synaptic locus and polygenic score for cognitive progression in Parkinson’s disease. Nat Genet. Jun 2021;53(6):787–793. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Tsuang D, Leverenz JB, Lopez OL, et al. APOE epsilon4 increases risk for dementia in pure synucleinopathies. JAMA Neurol. Feb 2013;70(2):223–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Miyamoto T, Miyamoto M. Odor identification predicts the transition of patients with isolated RBD: A retrospective study. Ann Clin Transl Neurol. Aug 2022;9(8):1177–1185. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Wolz M, Hahner A, Meixner L, et al. Accurate detection of Parkinson’s disease in tremor syndromes using olfactory testing. Eur Neurol. 2014;72(1–2):1–6. [DOI] [PubMed] [Google Scholar]
- 36.Roos DS, Klein M, Deeg DJH, et al. Prevalence of Prodromal Symptoms of Parkinson’s Disease in the Late Middle-Aged Population. J Parkinsons Dis. 2022;12(3):967–974. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Savica R, Grossardt BR, Bower JH, et al. Incidence of dementia with Lewy bodies and Parkinson disease dementia. JAMA Neurol. Nov 2013;70(11):1396–402. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Shrestha S, Zhu X, Griswold ME, et al. Olfaction and Plasma Biomarkers of Alzheimer Disease and Neurodegeneration in the Atherosclerosis Risk in Communities Study. Neurology. Jun 10 2025;104(11):e213706. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Ashton NJ, Brum WS, Di Molfetta G, et al. Diagnostic Accuracy of a Plasma Phosphorylated Tau 217 Immunoassay for Alzheimer Disease Pathology. JAMA Neurol. Mar 1 2024;81(3):255–263. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Mielke MM, Syrjanen JA, Blennow K, et al. Plasma and CSF neurofilament light: Relation to longitudinal neuroimaging and cognitive measures. Neurology. Jul 16 2019;93(3):e252–e260. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.