Discordance between amyloid-β PET and CSF biomarkers: Clinical and pathophysiological consequences

Among the earliest neuropathological events in Alzheimer’s disease (AD) is the accumulation and aggregation of amyloid-β, which occurs decades before symptom onset. Two methods, amyloid-β positron emission tomography (PET) and Aβ42 levels in the cerebrospinal fluid (CSF), are currently considered interchangeable for assessing amyloid-β pathology in vivo. However, in 10-20% of cases they show discordant (CSF+/PET- or CSF-/PET+) results. The causes for this discordance are largely unknown. In this thesis we used several different approaches, both in vivo and ex vivo, to investigate the possible independent information provided by amyloid-β PET and CSF. First, we investigated the clinical consequences of having discordant amyloid-β CSF/PET biomarker status. Then we focused on the potential differences of the CSF+/PET- and CSF-/PET+ participants by utilizing the predictive ability of various patient features. Next, we explored why sometimes amyloid-β PET is clinically requested for patients with available CSF biomarker analysis. Thereafter, we investigated longitudinal trajectories of amyloid-β accumulation, tau and cognition and studied whether amyloid-β CSF/PET discordant status is associated with tau 5 years later. Finally, we investigated the neuropathological underpinnings of amyloid-β CSF/PET concordance in a sample with available neuropathological data. We found that there was consistently ~15% discordance between amyloid-β PET and CSF in different samples and cohorts. The discordance rate differed by disease stage, being more frequent in early disease. Amyloid-β CSF+/PET- status was consistently more prevalent than CSF-/PET+. Although discordant amyloid-β status is not benign and is associated with amyloid-β pathology, it is associated with a distinctly better prognosis compared to having concordant positive amyloid-β biomarkers before dementia. Our results combined with previous work from others suggest that amyloid-β CSF/PET discordance is likely caused by different factors, both biological and methodological. We found that on a group level CSF+/PET- status is associated with early amyloid-β pathology where significant neuronal loss has not yet begun, offering opportunities for future trials. CSF+/PET- status could also be caused by other diseases, such as neuroinflammation, and methodological variation. Cases with CSF-/PET+ status are most often caused by a combination false-negative CSF in cases of AD, and false-positive PET signal in cases without AD. Although CSF and PET might not offer identical information about amyloid-β pathology, in the large majority of cases they are concordant and both can be used to assess in vivo amyloid-β pathology. In complicated clinical dilemmas, there even might be complementary value of using both biomarkers. In conclusion, our results offer insights to the pathophysiology of Alzheimer’s disease and the clinical utility of in vivo amyloid-β biomarkers.