A Diagnostic Index Based on MRI and CSF Biomarkers
CBN (Computational Biology and Neurocomputing) seminars
Friday 14 June 2013
to 11:00 at
Eric Westman (Department of Neurobiology, Care Sciences and Society, Karolinska Institutet, Stockholm)
Objective: Magnetic resonance imaging (MRI) and cerebrospinal fluid (CSF) biomarkers reflect different pathological processes in Alzheimer’s disease (AD). The aim of this study was to investigate a diagnostic index based on CSF and MRI biomarkers. Cognitive and biomarker patterns were analyzed in relation to the diagnostic index.
Method: 96 AD, 166 mild cognitive impairment (MCI) and 111 cognitively normal (CTL) subjects were included. A total of 60 variables were used for OPLS multivariate analysis (34 cortical thickness and 23 volumetric measures from Freesurfer and 3 CSF measures: Aβ42, t-tau and p-tau). An OPLS model was created combining CSF and MRI measures to distinguish between AD and CTL. An index was generated for each subject, close to one for AD and close to zero for CTL subjects. MCI subjects were predicted on to the AD/CTL model to investigate if they had a more AD-like or more CTL-like index.
Results: Correctly classified CTL subjects (mean index=0.10) had no abnormal biomarkers (Aβ42, t-tau, p-tau, hippocampus volume, entorhinal cortical thickness and CSF volume). Correctly classified AD subjects (mean index=0.91) had all biomarkers abnormal. CTL subjects with a more AD like index (mean index=0.65) had a clear AD like CSF profile, along with a thinner entorhinal cortex. Interestingly, CTL-like AD subjects (mean index=0.33) had only abnormal Aβ42 levels. The MCI subjects showed a more heterogonous profile (mean index MCIconverters= 0.66 and MCIstable=0.41).
Conclusion: An MRI/CSF based index is useful for early diagnosis, but also for targeting the right population for clinical trials.