Effect of A/T/N imaging biomarkers on impaired odor identification in Alzheimer's disease

Abstract

Odor identification ability may serve as an important diagnostic biomarker in Alzheimer's disease (AD). The aim of the study is to investigate the contribution of A/T/N neuroimaging biomarkers to impaired odor identification ability in the Alzheimer's disease spectrum. In 127 participants, we compared A/T/N neuroimaging biomarkers between normosmia and hyposmia groups, and performed correlation analysis between the biomarkers and Cross-Cultural Smell Identification Test (CCSIT) scores. Additionally, path analysis for odor identification ability was performed using cognitive function as a mediator. In between-group comparison, individuals with hyposmia showed higher frequency of amyloid-β (Aβ) positivity, and lower neuropsychological test performance than those with normosmia. After correction for covariates including total cognition scores, there was no difference in the Aβ or tau burden between the normosmia and hyposmia groups, and no correlation between CCSIT scores and Aβ or tau burden. Meanwhile, cortical volumes in the lateral and medial temporal cortices were smaller in the hyposmia group and decreased with the worsening of CCSIT scores. Path analysis showed that only neurodegeneration had a direct effect on odor identification, while Aβ and tau burden contributed to odor identification with the mediation of cognition. In the Alzheimer's disease spectrum, impaired odor identification ability may be attributable to neurodegeneration rather than the direct effect of Aβ or tau burden.

Figure 1

Correlation between the total cognition scores and CCSIT scores. Correlation analysis was performed after adjusting for age, sex, presence of ApoE ε4 and years of education in all 127 participants (a), and 59 amyloid-β positive participants (b). Blue dots represent CU, green dots MCI patients, and red dots DEM. Abbreviations: Aβ+ /− = Aβ-positivity, SUVR = standardized uptake value ratio, CU = cognitively unimpaired; MCI = mild cognitive impairment; DEM = dementia, ApoE = apolipoprotein E, SR = Standardized residual, CCSIT = Cross-Cultural Smell Identification Test.

Figure 2

Comparison of ¹⁸F-florbetaben, ¹⁸F-flortaucipir SUVR and cortical volume between normosmia and hyposmia groups in all participants (a), and amyloid-β positive participants (b). Regional SUVR values were compared between the two groups after adjusting for age, sex, years of education, presence of ApoE ε4 and total cognition scores. For the comparison of regional cortical volumes between the two groups, total intracranial volume was added to the list of covariates. Data are presented as means (horizontal bars) and standard deviations (error bars) of normosmia (red) and hyposmia (blue) groups. P-values for the difference between two groups are expressed as − log₁₀ P. P-values presented with red bars indicate the regions that survived correcting for region-wise multiple comparisons (false discovery rate-corrected P < 0.05), and blue dotted lines represent uncorrected P < 0.05. Abbreviations: Aβ+ /− = Aβ-positivity, SUVR = standardized uptake value ratio, A = ¹⁸F-florbetaben SUVR, T = ¹⁸F-flortaucipir SUVR, N = cortical volume.

Figure 3

Correlation of CCSIT score with ¹⁸F-florbetaben, ¹⁸F-flortaucipir SUVR and cortical volume in all participants (a), and amyloid-β positive participants (b). We used Pearson’s correlation between the standardized residuals of CCSIT scores and regional SUVR values obtained with multiple linear regression model after adjusting for age, sex, years of education, presence of ApoE ε4 and total cognition scores. A covariate of intracranial volume was additionally adjusted for cortical volume. Horizontal bars represent P-values as − log₁₀ P. Red bars represent the regions that survived correction for region-wise multiple comparisons (false discovery rate-corrected P < 0.05), and blue dotted lines represent uncorrected P < 0.05. In the right side panel, brain regions showing highest significance in correlation with ¹⁸F-florbetaben, ¹⁸F-flortaucipir, and cortical volume are presented respectively. Blue dots represent CU participants, green dots represent MCI patients, and red dots represent DEM. Abbreviations: Aβ+ /− = Aβ-positivity, SUVR = standardized uptake value ratio, A = ¹⁸F-florbetaben SUVR, T = ¹⁸F-flortaucipir SUVR, N = cortical volume, SR = Standardized residual, CCSIT = Cross-Cultural Smell Identification Test, CU = cognitively unimpaired; MCI = mild cognitive impairment; DEM = dementia, ApoE = apolipoprotein E.

Figure 4

Schematic diagram of the path analyses for effect of A/T/N imaging biomarkers on odor identification ability. Path analysis for odor identification ability was performed using ¹⁸F-florbetaben, ¹⁸F-flortaucipir SUVR and cortical volume as predictors and the total cognition score as a mediator. In the upper part, total effects (direct effect + indirect effect) of the A/T/N imaging biomarkers on odor identification are shown. In (c, d), the lower side of the triangle represents the direct effects of the cortical volume on odor identification ability. Indirect effects of the A/T/N imaging biomarkers on odor identification ability are obtained by multiplying two β-coefficients between A/T/N imaging biomarkers and cognition, and between cognition and odor identification ability. Only significant β-coefficients are shown. **P < 0.001, *P < 0.05.