Predicting Conversion from MCI to AD Combining Multi-Modality Data and Based on Molecular Subtype

Abstract

Alzheimer’s disease (AD) is a neurodegenerative brain disease in the elderly. Identifying patients with mild cognitive impairment (MCI) who are more likely to progress to AD is a key step in AD prevention. Recent studies have shown that AD is a heterogeneous disease. In this study, we propose a subtyping-based prediction strategy to predict the conversion from MCI to AD in three years according to MCI patient subtypes. Structural magnetic resonance imaging (sMRI) data and multi-omics data, including genotype data and gene expression profiling derived from peripheral blood samples, from 125 MCI patients were used in the Alzheimer’s Disease Neuroimaging Initiative (ADNI)-1 dataset and from 98 MCI patients in the ADNI-GO/2 dataset. A variational Bayes approximation model based on the multiple kernel learning method was constructed to predict whether an MCI patient will progress to AD within three years. In internal fivefold cross-validation within ADNI-1, we achieved an overall AUC of 0.83 (79.20% accuracy, 81.25% sensitivity, 77.92% specificity) compared to the model without subtyping, which achieved an AUC of 0.78 (76.00% accuracy, 77.08% sensitivity, 75.32% specificity). In external validation using ADNI-1 as a training set and ADNI-GO/2 as an independent test set, we attained an AUC of 0.78 (74.49% accuracy, 74.19% sensitivity, 74.63% specificity). Identifying MCI patient subtypes with omics data would improve the accuracy of predicting the conversion from MCI to AD. In addition to evaluating statistics, obtaining the significant sMRI, single nucleotide polymorphism (SNP) and mRNA expression data from peripheral blood of MCI patients is noninvasive and cost-effective for predicting conversion from MCI to AD.

Multiscale structural mapping of Alzheimer’s disease neurodegeneration

“…We also used the MSSM features to predict MCI patients who progress to AD dementia within 3 years and it showed prominent prognostic performance (AUROC = 0.908, AURPC = 0.910), which is comparable to but slightly higher than the cortical thickness-based model that used similar procedures and models based on hippocampal volume or RAVLT cognitive test scores. It also outperformed models that use one or more MRI features of cortical thickness, hippocampal volume, and volumetric features, ( Sørensen et al, 2016 , Popuri et al, 2020 , Zhu et al, 2017 ), MRI-based CNN models, ( Gao et al, 2020 , Lu et al, 2018 ), a model that combines MRI, genotypes, and gene expression profiles ( Li et al, 2021 ), a model combining MRI morphometrics, CSF and cognitive measures ( Ye et al, 2012 ), and models using PET or PET combined with MRI, ( Liu et al, 2014 , Lu et al, 2018 , Zhu et al, 2017 ). Higher performance was reported in a few studies that used longitudinal MRI, ( Sun et al, 2017 ) and that combined MRI with cognitive performance ( Tong et al, 2017 ).…”

“…Neurodegeneration measures from imaging are associated with disease severity and prognosis ( Benvenutto et al, 2018 , Brickman et al, 2018 , Aziz et al, 2017 ), and MRI is an alternative tool to assess AD neurodegeneration, which is safe, non-invasive, and more accessible in clinical settings. Advances in brain imaging with machine learning techniques made it possible to identify individuals with AD dementia or mild cognitive impairment (MCI) ( Belathur Suresh et al, 2018 , Cho et al, 2012 , Suk et al, 2014 , Sarraf et al, 2019 , Bloch and Friedrich, 2021 , Lin et al, 2018 , Li et al, 2021 , Guo et al, 2017 , Ye et al, 2012 , Gao et al, 2020 , Shi et al, 2018 , Liu et al, 2014 , Wolz et al, 2011 , Lu et al, 2018 , Sørensen et al, 2016 , Popuri et al, 2020 , Sun et al, 2017 , Tong et al, 2017 , Allison et al, 2019 , Noor et al, 2019 , Choi et al, 2020 , Zhang et al, 2011 , Westman et al, 2012 , Park et al, 2017 , Davatzikos et al, 2008 , Desikan et al, 2009 , Sørensen et al, 2017 , Zhu et al, 2017 , McEvoy et al, 2009 , Magnin et al, 2009 , Mattsson et al, 2019 , Janghel and Rathore, 2021 ) using fairly routine structural MRI procedures and therefore MRI provides an optimal ‘first pass’ screen of patients. A simple metric such as hippocampal volume is sensitive to AD neurodegeneration; however hippocampal volume is impacted by a range of conditions and is not specific to AD.…”

The promise of multi-omics approaches to discover biological alterations with clinical relevance in Alzheimer’s disease

“…This suggests that MAPKs are the principal modulators of the cognitive impairment and decline independently of the APOE genotype. According to Li et al (2021) , conversion risk differs between patient groups presenting distinct genomic, proteomic and neuroimaging features. Using proteomics, genomics and transcriptomics data derived from brain tissue, Seyfried et al identified a genetic risk locus for AD encoding genes involved in microglia and oligodendrocyte function associated with cognitive decline ( Seyfried et al, 2017 ).…”

Long-Term Neuromodulatory Effects of Repetitive Transcranial Magnetic Stimulation (rTMS) on Plasmatic Matrix Metalloproteinases (MMPs) Levels and Visuospatial Abilities in Mild Cognitive Impairment (MCI)

“…Mild cognitive impairment (MCI) is a neurological disorder occurring mainly in older people with cognitive deficits that are not severe enough to warrant a diagnosis of dementia [ 1 ]. It has been classified as a prodromal stage of a variety of dementing disorders, including Alzheimer’s disease (AD) [ 2 ].…”