Premium
Biomarker profiles for staging of pre‐dementia patients
Author(s) -
Kuehnel Line,
Bouteloup Vincent,
Lespinasse Jérémie,
Chêne Geneviève,
Dufouil Carole,
Raket Lars Lau
Publication year - 2020
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1002/alz.042966
Subject(s) - biomarker , disease , oncology , dementia , stage (stratigraphy) , medicine , imaging biomarker , psychology , magnetic resonance imaging , radiology , biology , paleontology , biochemistry , chemistry
Background Patient cohorts in clinical trials for Alzheimer’s disease are often highly inhomogeneous affecting statistical power to detect treatment effects. Objective methods that enable better staging of patients can be used to design inclusion criteria that result in less variability in progression and hereby increase statistical power. Furthermore, better prediction of disease stage is a key issue to determine when an intervention may have highest chance of success. Method Continuous time‐progression of cognitive decline was estimated via a nonlinear mixed‐effects disease‐progression model. The model simultaneously estimated evolution of MMSE scores over the course of the disease and predicted patient disease stage. To investigate if combinations of baseline CSF and imaging biomarkers were predictive of disease stage, AIC‐based forward selection was used to include biomarker effects on disease stage in the disease‐progression model. In addition to biomarkers, effects of education, sex and age were also investigated. The included biomarkers were MRI cortical thickness signature, FDG PET, and core CSF biomarkers (total tau, phosphorylated tau, aβ42, aβ40, and aβ42/aβ40 ratio). To investigate the utility of biomarker profile on the model’s ability to predict disease stage, its performance in predicting future progression on MMSE was compared to a model that did not include biomarkers. The model was fitted to individuals with subjective or mild cognitive impairment (SCI/MCI) from the French MEMENTO study. The results of the model were validated in individuals with either significant memory complaints or MCI in ADNI. Result There were 312 individuals with SCI/MCI and complete biomarkers available at baseline. The optimal model included effects of total tau, MRI cortical thickness, and aβ42 on patients’ disease stage measured via MMSE scores. A model with the same terms was fitted to data from 609 individuals from ADNI to validate the findings. Inclusion of selected biomarkers was shown to improve staging of patients and to have similar directional contributions as in MEMENTO. Conclusion We demonstrated that biomarker profiles are predictive of disease stage in pre‐dementia. We found that a combination of MRI cortical thickness, CSF total tau and CSF aβ42 was predictive of disease stage in the French MEMENTO cohort and validated this finding in ADNI.