Performance on the Brief Smell Identification test ranged from 0 to 12 correct (mean = 8.0, SD = 2.6). Scores were inversely related to age at death (estimate = −0.10, standard error [SE] = 0.03, P < 0.001) but unrelated to sex (estimate = −0.033, SE = 0.32, P = 0.31) and education (estimate = −0.09, SE = 0.06, P = 0.11).
Odor identification and Lewy bodies
On neuropathological examination, Lewy bodies were identified in 26 individuals (12.9%) and classified as nigral predominant (i.e., restricted to substantia nigra) in 4, limbic type (i.e., in both nigral and limbic regions) in 9, and diffuse neocortical type (i.e., in nigral, limbic, and neocortical regions) in 13. Those with Lewy bodies did not demographically differ from unaffected persons, but they had lower olfactory test scores, higher likelihood of developing dementia, more parkinsonian motor dysfunction, and higher levels of AD pathology and nigral cell loss ().
Descriptive information on persons with versus without Lewy bodies on neuropathological examinationa
We examined the relation of Lewy bodies to olfactory performance in a series of linear regression models that controlled for age at death, sex, education, and time from olfactory testing to death. In an initial analysis, the presence of any Lewy bodies was associated with a loss of about 3 points on the olfactory test (estimate = −3.03, SE = 0.47, P < 0.001) and accounted for 15.4% of the variance on the test.
To determine whether the association of Lewy bodies with olfaction depended on their distribution, we repeated the analysis with separate indicators for the 3 Lewy body subgroups. As shown in model A of , there was no olfactory impairment in the nigral predominant Lewy body subgroup but substantial olfactory dysfunction in the other 2 Lewy body subgroups. Lewy bodies accounted for 19.1% of the variance in olfactory performance in this model. shows the model-based estimates of the mean olfactory score and 95% confidence interval for each Lewy body subgroup in comparison to unaffected persons. The figure suggests decreased olfactory function with increased severity of Lewy body disease, with no impairment in the nigral subgroup, moderate impairment in the limbic subgroup, and profound impairment in the neocortical subgroup.
Relation of postmortem pathological measures to odor identificationa
Odor identification in Lewy body subgroups. Estimated mean and 95% confidence intervals on odor identification test in Lewy body subgroups, adjusted for age at death, sex, education, and time from olfactory testing to death.
Cognitive impairment is associated with both Lewy bodies (Wilson et al. 2010
) and olfactory identification (Wilson et al. 2006
). Therefore, we repeated the previous analysis with a term to control for the presence of dementia at the time of olfactory testing (n
= 28). Both limbic and neocortical Lewy bodies continued to be robustly related to olfactory performance (, model B). As shown in model C of , results were similar when we repeated the analysis with a term to control for dementia at any time during the study (n
To determine whether preclinical Parkinson's disease was affecting results, we repeated the core analysis with a term added for parkinsonian motor dysfunction at the time of olfactory testing, as assessed with a modified version of the Unified Parkinson's Disease Rating Scale. The association of Lewy body subgroups with olfactory score was not substantially changed in this analysis or in a subsequent model that controlled for parkinsonian score most proximate to death (data not shown).
Odor identification and AD pathology
To see if AD pathology affected the association of Lewy bodies with olfactory functioning, we conducted further analyses using a composite measure of diffuse plaques, neuritic plaques, and neurofibrillary tangles. Pathologic scores ranged from 0 to 3.2 (mean = 0.62, SD = 0.61). In an initial model, higher level of AD pathology was associated with lower olfactory score and accounted for 8.4% of the variability in performance (estimate = −1.25, SE = 0.27, P < 0.001). When terms for Lewy body subgroups were added to the model, the effect of Lewy bodies was undiminished and AD pathology accounted for an additional 4.1% of performance variability (, model D).
Odor identification and nigral neuronal loss
We conducted additional analyses to determine whether another neuropathologic hallmark of Parkinson's disease, neuronal loss in the substantia nigra, affected olfaction or its association with Lewy bodies. The neuropathologist's ratings of nigral cell loss ranged from 0 (none) to 5 (severe), had an approximately normal distribution (mean = 2.33, SD = 0.94, skewness = 0.41), and were higher in those with Lewy bodies (Spearman rho = 0.31, P < 0.001). In a regression analysis, higher level of nigral cell loss was associated with lower olfactory score (estimate = 0.71, SE = 0.19, P < 0.001). However, the effect was reduced by more than half and no longer significant when terms for Lewy bodies were added to the analysis (, model E).
Detection of underlying Lewy body disease
Logistic regression analyses were used to assess the value of olfactory test performance in identifying individuals with underlying Lewy body disease (). Each model provided an estimate of the area under the receiver operating characteristic curve, an overall index of how well affected individuals were identified. The first analysis included a core set of 4 covariates (i.e., age at death, sex, education, and time from olfactory testing to death). As shown in (model A), the area under the receiver operating curve for this model was 0.62. When the analysis was repeated with a term added for odor identification score, the area under the curve was 0.82. , which is based on these analyses, shows a substantial increase in predictive accuracy in the model with olfactory data (red line) compared with the reference model without olfactory data (blue line) and chance (dashed line). Results were comparable when the analyses were repeated controlling for dementia at the time of olfactory testing (, model B). The relation of olfactory score to postmortem Lewy bodies was similar when the analysis adjusted for dementia at any point during the study except that the increase in predictive accuracy was not quite significant because of improved prediction in the reference model (, model C).
Relation of odor identification score to postmortem evidence of Lewy bodiesa
Figure 2 Use of olfactory data to detect Lewy body disease. Estimated area under the receiver operating characteristic curve for models with (red line) and without (blue line) odor identification score, adjusted for age at death, sex, education, and time from (more ...)