We showed that plasma MDS-OAβ has good accuracy to pre-screen for brain amyloidosis in a memory clinic population, particularly when combined with APOEe4 and age (AUC > 0.80). In addition, plasma MDS-OAβ showed a negative correlation with CSF Aβ42 and MMSE, and a positive correlation with CSF t-Tau. Using plasma MDS-OAβ as a pre-screener resulted in reduced number of PET scans and lowered costs for amyloid screening up to 40%, which is highly beneficial for clinical trials. Of note, these results are only valid for relatively fresh samples, as a negative effect of long-term storage was found for plasma MDS-OAβ concentrations.
To date, few studies have measured Aβ oligomers levels in blood plasma, as detecting crude oligomeric Aβ in plasma is challenging owing to its low concentration in blood. Using the MDS platform, we successfully measured increased plasma MDS-OAβ levels in abnormal amyloid-PET individuals compared to individuals with normal amyloid-PET levels. This finding is in line with previous studies reporting increased levels of Aβ oligomers in brain tissue, CSF, and plasma of AD patients. (Georganopoulou et al., 2005; Santos et al., 2008; Xia et al., 2009; Zhou et al., 2012) This increase in Aβ oligomer levels is in contrast to monomeric Aβ levels, which show an evident decrease rather than increase in blood plasma. (Janelidze et al., 2016; Ovod et al., 2017; Nakamura et al., 2018; Verberk et al., 2018; Palmqvist et al., 2019) This upregulation of Aβ oligomers could be explained by oligomerization of Aβ monomers, resulting in higher plasma Aβ oligomer levels and decreased monomeric Aβ levels. Our results also showed a correlation between plasma MDS-OAβ and CSF Aβ42, t-Tau or MMSE scores, which is in line with previous plasma Aβ monomer studies. (Janelidze et al., 2016; An et al., 2017; Verberk et al., 2018; Palmqvist et al., 2019) However, these correlations were not strong, and an explanation for this could be the peripheral production of plasma Aβ, by platelets, skeletal muscle cells and other cell types (Roher et al., 2009) that contribute to circulating Aβ levels resulting in a dilution of the relation with CNS processes.
This is the first study to report on plasma AβOs as a marker for brain amyloidosis in a large amyloid PET confirmed cohort. As the definition of AD is shifting from a syndrome to a biological construct, it is relevant to evaluate the performance of biomarkers in discriminating amyloid status (Jack et al., 2018). One small-scale study did evaluate oligomeric assemblies of misfolded Aβ protein as a plasma marker for amyloid status between prodromal PET-positive individuals (n=36) and healthy elderly PET-negative individuals (n=37) (Nabers et al., 2018). Using an immune-infrared sensor method they achieved an AUC of 0.78 (95% CI 0.68–0.88) (Nabers et al., 2018). We showed a similar good accuracy of plasma MDS-OAβ to screen for amyloid status (AUC: 0.81) in a large amyloid PET confirmed cohort when combined with APOEe4 and age. When restricting the analyses to individuals in pre-dementia stages (i.e. CN and MCI), the accuracy of plasma MDS-OAβ combined with APOEe4 and age increased further to 0.86.
It is well known that preanalytical factors concerning sample handling and processing can influence the measured concentration of (plasma) biomarkers, therefore leading to variability in results, preventing establishment of a universal cutoffs and between-laboratory comparisons (Hansson et al., 2018; Rózga et al., 2019). We previously observed a negative effect of long-term storage on plasma MDS-OAβ levels upon visual inspection (data not published), and therefore, decided to evaluate this in the current study. We found that with longer shelf-life plasma MDS-OAβ levels no longer differed between normal and abnormal amyloid individuals. Our finding is not fully in line with one recent study that investigated the long term storage effect on plasma monomeric Aβ, and found stable plasma Aβ levels after long-term storage up to 5 years at -80°C (Chiu et al., 2019). This discrepancy might be caused by the difference in storage length between the previous study (up to 5 years) and the current study (up to 19 years). It might also be caused by the difference in analytical methods (MDS vs. IMR) or the different Aβ species (MDS-OAβ vs (in principle) monomeric Aβ42). It could be the case that plasma MDS-OAβ levels of normal amyloid increase over time and reach similar levels as plasma MDS-OAβ levels of abnormal amyloid individuals, whereas in monomeric Aβ42 this does not happen. It could be hypothesized that long-term storage might induce stress on the oligomeric Aβ42 protein which results in perturbation of the protein and an increased aggregation tendency in normal amyloid individuals, which does not occur in abnormal amyloid individuals as they have already reached maximum oligomerization. A similar increase in protein aggregation induced by protein-stress has previously been reported after freeze-thawing (Carpenter, Manning and Randolph, 2002). The effect of long-term storage time implies that the plasma MDS-OAβ assay cannot be used to perform research projects with samples that have been stored in biobanks for a long period. This novel finding could be of interest to other research groups interested in measuring plasma biomarkers of amyloid. Additional pre-analytical testing is needed to determine the precise maximum storage period, and to compare the effect of long-term storage with other types of blood based Aβ biomarkers. Nonetheless, as in daily clinical routine fresh blood samples are used we do not expect this will present a problem for daily clinical practice.
Strengths and limitations
Our study has several strengths including our large well-defined amyloid PET confirmed memory clinic population. In addition, CSF and plasma collection follows a highly standardized protocol in our center, thus minimizing confounding effects in pre-analytical processing. Moreover, the oligomerization assay technique developed for this plasma MDS-OAβ assay can potentially be employed for other proteinopathies as well, such as α-synuclein which is often seen in dementia with Lewy bodies. This might result in a screening panel of plasma biomarkers for different types of neurodegenerative disease. Among the limitations of our study is that the plasma MDS-OAβ assay is not yet available on an automatic platform, thus enhancing the risk for analytical variation. However, automation is currently under development, further facilitating broad implementation and minimizing analytical variation. In addition, plasma MDS assays for other AD biomarkers, such as phosphorylated Tau, are currently under development to further capture the full pathological profile of AD (Jack et al., 2018). Lastly, it would be interesting to study the association between plasma MDS-OAβ with specific cognitive domains, through elaborate neuropsychological testing, to get an in-depth understanding of the association between plasma MDS-OAβ and cognitive impairment.