Student thesis: Doctoral ThesisDoctor of Philosophy


Alzheimer´s disease (AD) biomarkers that can detect and track disease progression at its earliest stages to aid the critical search for a disease modifying therapy is much needed. Markers of in vivo amyloid-beta (Aβ) deposition (e.g. 11C-PiB) combined with positron emission tomography (PET) or cerebrospinal fluid (CSF) examination are becoming widely utilised as essential criterion for AD prevention trials. Although necessary, this is likely to come at a great cost and will restrict the progression of some trials. The inexpensive and accessible nature of a blood-based prediction for AD risk would be of considerable value in a population screening process. The traditional “case versus control” design frequently disregards the clinical heterogeneity in AD, with active preclinical neuropathology overlooked. Therefore, deriving biologically relevant markers associated with in vivo surrogates of AD pathology is considered a superior approach. Here, we aimed to identify single and multi-analyte plasma biomarkers associated with neocortical Aβ burden (NAB) using two proteomic approaches.
One dimensional gel electrophoresis (1DGE) coupled with Mass Spectrometry was performed on 78 individuals with extreme ranges of NAB. Immunoassay-based techniques were utilised to validate protein candidates in independent cohorts. Further to this, an improved proteomic strategy incorporating high-resolution peptide separation was able to increase the number of quantifiable targets and widen plasma proteome coverage. This enhanced methodology was applied to 297 individuals from two cohorts stratified by Aβ PET. In both discoveries, the relationship between plasma protein levels and NAB modalities was examined, with an attempt to build multi-modal predictions of NAB.
In the first discovery study, several candidates associated with NAB were selected for technical replication. Plasma FGγ models predicted NAB with a sensitivity of 59% and specificity of 78%. FGγ was further shown to associate with Aβ using core CSF biomarkers as surrogate measures. The secondary discovery study demonstrated a larger number of single markers associated with NAB, along with the verification of brain-derived proteins found to be present in plasma. A machine learning analysis built a multi-analyte panel for NAB prediction which was shown to replicate, in an independent cognitively normal cohort, with an accuracy of 86.6%. This predictive panel indicated the convergence of pathways related to coagulation, APP processing, neuronal transcription factors and axonal injury to be of central importance in predicting NAB.
Date of Award2017
Original languageEnglish
Awarding Institution
  • King's College London
SupervisorAbdul Hye (Supervisor) & John Powell (Supervisor)

Cite this