A Proteomics Approach to the Diagnosis of Alzheimer's Disease

Alzheimer's disease (AD) is the most common form of dementia in the elderly and affects almost 30% of the population over 85 years of age. The cause and mechanisms of the disease are still unknown, yet a large body of evidence supports the so called amyloid cascade hypothesis. According to this model, abnormal accumulation and deposition of Aβ peptide lead through a series of processes to fibril formation and synaptic dysfunction, neuronal loss and finally dementia. No cure for this disease is available yet and definite diagnosis is only possible by histological analysis of post mortem brain tissue. The diagnosis of a patient is a very complicated, expensive and time-consuming process including a variety of analyses such as neuropsychological testing, neuro-imaging, and blood tests. In specialized centers the accuracy of this diagnostic workup reaches 80–90%. The aim of this study was to screen the proteome of the cerebrospinal fluid (CSF) of patients with AD, patients with other neurological disorders including other types of dementia, and healthy age-matched control subjects for differences. For the CSF proteome analysis surface enhanced laser desorption ionization time-of-flight mass spectrometry (SELDI-TOF MS) was used. This method combines a chromatographic sample prefractionation step and mass spectrometric protein analysis in one technology. The data was analyzed using classification tree software, a bio-informatics tool based on a binary recursive partitioning algorithm. A proteomic pattern for the diagnosis of Alzheimer's disease was found consisting of 5 nodes representing peaks with molecular weights of 14 kDa, 11.7 kDa, 66 kDa, 4.2 kDa, and 80 kDa. This pattern reached a sensitivity of 80% and specificities of 75% (AD versus healthy control subjects), 95% (AD versus other neurological disorders), and 85% (AD versus both other groups). These values are comparable to the diagnostic accuracy obtained by standard diagnostic procedures. However, the combination of several diagnostic patterns into a "committee of experts" analysis led to correct classification of up to 100% of the test samples. By these patterns together with univariate statistical analysis of the proteomic data seven putative bio-marker candidates were found, three of which were identified. Two of these marker candidates were shown to be different forms of transthyretin, namely the glutathionylated form and an N-terminally truncated form, whereas the third bio-marker was shown to be Apolipoprotein AI. Transthyretin is a thyroid hormone carrier protein which also binds Aβ peptide and has been reported to protect neurons against Aβ toxicity. Apolipoprotein AI is the major component of plasma HDL and participates in the reverse transport of cholesterol from tissues to the liver. Apolipoprotein AI therefore plays a role in cholesterol metabolism and homeostasis which is believed to play a role in the mechanism of Alzheimer's disease pathology. The putative markers not yet identified have molecular weights of 4.6 kDa, 7.7 kDa, 40 kDa and 51 kDa and seem to separate more generally the healthy status from neurological disease. In conclusion, the principle of using protein patterns in CSF for the diagnosis of Alzheimer's disease was demonstrated and three bio-marker candidates which may be interesting from a pathophysiological point of view were identified.