Salivary levels of amyloid beta reflect brain amyloid beta burden in cognitively-normal older adults.

Background: Amyloid beta (Aβ) plaque burden, as measured by positron emission tomography (PET), is increasingly being used as a biomarker for Alzheimer's disease (AD) as well as a screening or monitoring tool for clinical trials with amyloid-lowering drugs. However, PET imaging is expensive, invasive and not widely available for all patients, necessitating alternative means to assess brain Aβ accumulation.

Objectives: In this study, we measured levels of Aβ42, Aβ40 and Aβ38 in saliva samples from cognitively unimpaired older adults (n=93; 61.7 % female; mean age = 70.1 ± 6.6 years) using the Mesoscale Discovery platform, carefully considering preanalytical variables, including timing of sample collection, blood contamination and sample concentration. We next determined the relationships between Aβ peptide levels and Aβ plaque burden within the brain, determined using 18F-florbetapir (FBP) PET.

Results: We found that salivary levels of Aβ38 and Aβ42, but not Aβ40 nor the Aβ42/Aβ40, were significantly positively correlated with the global mean FBP standardized uptake value ratio (SUVR), before and after adjusting for age, sex and time of day of saliva sample collection (r=0.523/0.544, p=0.001/0.002 and r=0.316/0.32, p=0.031/0.044, for Aβ38 and Aβ42, respectively). Similar results were observed when Aβ values were analyzed as a ratio to the total protein levels in each sample and when tested in saliva samples that were collected during a restricted morning time window. Using composite regions which represent cortical regions vulnerable to Aβ accumulation in early, intermediate, and late stages of AD, we found that Aβ38 showed the most robust correlation with FBP SUVRs from early-accumulating brain regions (r=0.510; p<0.001). In contrast to the observed effects in saliva, plasma levels of Aβ42 measured from a subset of the participants showed a significant negative correlation to mean FBP SUVR. Using logistic regression analysis to determine whether any salivary Aβ species could predict brain Aβ burden, we found that salivary levels of Aβ38 in combination with age, sex, sample timing and APOE genotype could predict Aβ-PET positivity with an area under the curve = 0.950 (95 % confidence interval, 0.876-1.0; p<0.0001).

Conclusions: Our findings suggest that salivary Aβ38 and/or Aβ42 could have relevance as a non-invasive, and more widely applicable biomarker, for utility in clinical studies on AD.