A Community-Based Study Identifying Metabolic Biomarkers of Mild Cognitive Impairment and Alzheimer's Disease Using Artificial Intelligence and Machine Learning

NMR-based metabolomics in Alzheimer's disease research: a review

Alzheimer's disease (AD) is a progressive neurodegenerative disorder and represents the most common cause of dementia in the elderly population worldwide. Currently, there is no cure for AD, and the continuous increase in the number of susceptible individuals poses one of the most significant emerging threats to public health. However, the molecular pathways involved in the onset and progression of AD are not fully understood. This information is crucial for developing less invasive diagnostic instruments and discovering novel potential therapeutic targets. Metabolomics studies the complete ensemble of endogenous and exogenous metabolites present in biological specimens and may provide an interesting approach to identify alterations in multiple biochemical processes associated with AD onset and evolution. In this mini review, we summarize the results from metabolomic studies conducted using nuclear magnetic resonance (NMR) spectroscopy on human biological samples (blood derivatives, cerebrospinal fluid, urine, saliva, and tissues) from AD patients. We describe the metabolic alterations identified in AD patients compared to controls and to patients diagnosed with mild cognitive impairment (MCI). Moreover, we discuss the challenges and issues associated with the application of NMR-based metabolomics in the context of AD research.

Non-invasive biomarkers for mild cognitive impairment and Alzheimer's disease

Alzheimer's disease is the most common type of dementia in the elderly. It is a progressive degenerative disorder that may begin to develop up to 15 years before clinical symptoms appear. The identification of early biomarkers is crucial to enable a prompt diagnosis and to start effective interventions. In this work, we conducted a metabolomic study using proton Nuclear Magnetic Resonance (1H NMR) spectroscopy in serum samples from patients with neuropathologically confirmed Alzheimer's disease (AD, n = 51), mild cognitive impairment (MCI, n = 27), and cognitively healthy controls (HC, n = 50) to search for metabolites that could be used as biomarkers. Patients and controls underwent yearly clinical follow-ups for up to six years. MCI group included samples from three subgroups of subjects with different disease progression rates. The first subgroup included subjects that remained clinically stable at the MCI stage during the period of study (stable MCI, S-MCI, n = 9). The second subgroup accounted for subjects which were diagnosed with MCI at the moment of blood extraction, but progressed to clinical dementia in subsequent years (MCI-to-dementia, MCI-D, n = 14). The last subgroup was composed of subjects that had been diagnosed as dementia for the first time at the moment of sample collection (incipient dementia, Incp-D, n = 4). Partial Least Square Discriminant Analysis (PLS-DA) models were developed. Three models were obtained, one to discriminate between AD and HC samples with high sensitivity (93.75%) and specificity (94.75%), another model to discriminate between AD and MCI samples (100% sensitivity and 82.35% specificity), and a last model to discriminate HC and MCI with lower sensitivity and specificity (67% and 50%). Differences within the MCI group were further studied in an attempt to determine those MCI subjects that could develop AD-type dementia in the future. The relative concentration of metabolites, and metabolic pathways were studied. Alterations in the pathways of alanine, aspartate and glutamate metabolism, pantothenate and CoA biosynthesis, and beta-alanine metabolism, were found when HC and MCI- D patients were compared. In contrast, no pathway was found disturbed in the comparison of S-MCI with HC groups. These results highlight the potential of 1H NMR metabolomics to support the diagnosis of dementia in a less invasive way, and set a starting point for the study of potential biomarkers to identify MCI or HC subjects at risk of developing AD in the future.

Metabolomic profiling of CSF and blood serum elucidates general and sex-specific patterns for mild cognitive impairment and Alzheimer's disease patients

Background

Beta-amyloid (Abeta) and tau protein in cerebrospinal fluid (CSF) are established diagnostic biomarkers for Alzheimer's disease (AD). However, these biomarkers may not the only ones existing parameters that reflect Alzheimer's disease neuropathological change. The use of quantitative metabolomics approach could provide novel insights into dementia progression and identify key metabolic alterations in CSF and serum.

Methods

In the present study, we quantified a set of 45 metabolites in CSF (71 patients) and 27 in serum (76 patients) in patients with mild cognitive impairment (MCI), AD, and controls using nuclear magnetic resonance (NMR)-based metabolomics.

Results

We found significantly reduced CSF (1.32-fold, p = 0.0195) and serum (1.47-fold, p = 0.0484) levels of the ketone body acetoacetate in AD and MCI patients. Additionally, we found decreased levels (1.20-fold, p = 0.0438) of the branched-chain amino acid (BCAA) valine in the CSF of AD patients with increased valine degradation pathway metabolites (such as 3-hydroxyisobutyrate and α-ketoisovalerate). Moreover, we discovered that CSF 2-hydroxybutyrate is dramatically reduced in the MCI patient group (1.23-fold, p = 0.039). On the other hand, vitamin C (ascorbate) was significantly raised in CSF of these patients (p = 0.008). We also identified altered CSF protein content, 1,5-anhydrosorbitol and fructose as further metabolic shifts distinguishing AD from MCI. Significantly decreased serum levels of the amino acid ornithine were seen in the AD dementia group when compared to healthy controls (1.36-fold, p = 0.011). When investigating the effect of sex, we found for AD males the sign of decreased 2-hydroxybutyrate and acetoacetate in CSF while for AD females increased serum creatinine was identified.

Conclusion

Quantitative NMR metabolomics of CSF and serum was able to efficiently identify metabolic changes associated with dementia groups of MCI and AD patients. Further, we showed strong correlations between these changes and well-established metabolomic and clinical indicators like Abeta.

Machine Learning Selection of Most Predictive Brain Proteins Suggests Role of Sugar Metabolism in Alzheimer's Disease

BACKGROUND

The complex and not yet fully understood etiology of Alzheimer's disease (AD) shows important proteopathic signs which are unlikely to be linked to a single protein. However, protein subsets from deep proteomic datasets can be useful in stratifying patient risk, identifying stage dependent disease markers, and suggesting possible disease mechanisms.

OBJECTIVE

The objective was to identify protein subsets that best classify subjects into control, asymptomatic Alzheimer's disease (AsymAD), and AD.

METHODS

Data comprised 6 cohorts; 620 subjects; 3,334 proteins. Brain tissue-derived predictive protein subsets for classifying AD, AsymAD, or control were identified and validated with label-free quantification and machine learning.

RESULTS

A 29-protein subset accurately classified AD (AUC = 0.94). However, an 88-protein subset best predicted AsymAD (AUC = 0.92) or Control (AUC = 0.92) from AD (AUC = 0.98). AD versus Control: APP, DHX15, NRXN1, PBXIP1, RABEP1, STOM, and VGF. AD versus AsymAD: ALDH1A1, BDH2, C4A, FABP7, GABBR2, GNAI3, PBXIP1, and PRKAR1B. AsymAD versus Control: APP, C4A, DMXL1, EXOC2, PITPNB, RABEP1, and VGF. Additional predictors: DNAJA3, PTBP2, SLC30A9, VAT1L, CROCC, PNP, SNCB, ENPP6, HAPLN2, PSMD4, and CMAS.

CONCLUSION

Biomarkers were dynamically separable across disease stages. Predictive proteins were significantly enriched to sugar metabolism.

Systematic Review of NMR-Based Metabolomics Practices in Human Disease Research

Nuclear magnetic resonance (NMR) spectroscopy is one of the principal analytical techniques for metabolomics. It has the advantages of minimal sample preparation and high reproducibility, making it an ideal technique for generating large amounts of metabolomics data for biobanks and large-scale studies. Metabolomics is a popular “omics” technology and has established itself as a comprehensive exploratory biomarker tool; however, it has yet to reach its collaborative potential in data collation due to the lack of standardisation of the metabolomics workflow seen across small-scale studies. This systematic review compiles the different NMR metabolomics methods used for serum, plasma, and urine studies, from sample collection to data analysis, that were most popularly employed over a two-year period in 2019 and 2020. It also outlines how these methods influence the raw data and the downstream interpretations, and the importance of reporting for reproducibility and result validation. This review can act as a valuable summary of NMR metabolomic workflows that are actively used in human biofluid research and will help guide the workflow choice for future research.

Metabolomic alterations in the blood plasma of older adults with mild cognitive impairment and Alzheimer's disease (from the Nakayama Study)

Alzheimer's disease (AD) is a progressive disease, and the number of AD patients is increasing every year as the population ages. One of the pathophysiological mechanisms of AD is thought to be the effect of metabolomic abnormalities. There have been several studies of metabolomic abnormalities of AD, and new biomarkers are being investigated. Metabolomic studies have been attracting attention, and the aim of this study was to identify metabolomic biomarkers associated with AD and mild cognitive impairment (MCI). Of the 927 participants in the Nakayama Study conducted in Iyo City, Ehime Prefecture, 106 were selected for this study as Control (n = 40), MCI (n = 26), and AD (n = 40) groups, matched by age and sex. Metabolomic comparisons were made across the three groups. Then, correlations between metabolites and clinical symptoms were examined. The blood mRNA levels of the ornithine metabolic enzymes were also measured. Of the plasma metabolites, significant differences were found in ornithine, uracil, and lysine. Ornithine was significantly decreased in the AD group compared to the Control and MCI groups (Control vs. AD: 97.2 vs. 77.4; P = 0.01, MCI vs. AD: 92.5 vs. 77.4; P = 0.02). Uracil and lysine were also significantly decreased in the AD group compared to the Control group (uracil, Control vs. AD: 272 vs. 235; P = 0.04, lysine, Control vs. AD: 208 vs. 176; P = 0.03). In the total sample, the MMSE score was significantly correlated with lysine, ornithine, thymine, and uracil. The Barthel index score was significantly correlated with lysine. The instrumental activities of daily living (IADL) score were significantly correlated with lysine, betaine, creatine, and thymine. In the ornithine metabolism pathway, the spermine synthase mRNA level was significantly decreased in AD. Ornithine was decreased, and mRNA expressions related to its metabolism were changed in the AD group compared to the Control and MCI groups, suggesting an association between abnormal ornithine metabolism and AD. Increased betaine and decreased methionine may also have the potential to serve as markers of higher IADL in elderly persons. Plasma metabolites may be useful for predicting the progression of AD.

Alteration of plasma metabolic profile and physical performance combined with metabolites is more sensitive to early screening for mild cognitive impairment

Objective

Unbiased metabolic profiling has been initiated to identify novel metabolites. However, it remains a challenge to define reliable biomarkers for rapid and accurate diagnosis of mild cognitive impairment (MCI). Our study aimed to evaluate the association of serum metabolites with MCI, attempting to find new biomarkers and combination models that are distinct for MCI.

Methods

A total of 380 participants were recruited (mean age: 72.5 ± 5.19 years). We performed an untargeted metabolomics analysis on older adults who underwent the Mini-Mental State Examination (MMSE), the Instrumental Activities of Daily Living (IADL), and physical performance tests such as hand grip, Timed Up and Go Test (TUGT), and walking speed. Orthogonal partial least squares discriminant analysis (OPLS-DA) and heat map were utilized to distinguish the metabolites that differ between groups.

Results

Among all the subjects, 47 subjects were diagnosed with MCI, and methods based on the propensity score are used to match the MCI group with the normal control (NC) group (n = 47). The final analytic sample comprised 94 participants (mean age: 75.2 years). The data process from the metabolic profiles identified 1,008 metabolites. A cluster and pathway enrichment analysis showed that sphingolipid metabolism is involved in the development of MCI. Combination of metabolite panel and physical performance were significantly increased discriminating abilities on MCI than a single physical performance test [model 1: the area under the curve (AUC) = 0.863; model 2: AUC = 0.886; and model 3: AUC = 0.870, P < 0.001].

Conclusion

In our study, untargeted metabolomics was used to detect the disturbance of metabolism that occurs in MCI. Physical performance tests combined with phosphatidylcholines (PCs) showed good utility in discriminating between NC and MCI, which is meaningful for the early diagnosis of MCI.

Machine Learning Model Based on Lipidomic Profile Information to Predict Sudden Infant Death Syndrome

Sudden infant death syndrome (SIDS) represents the leading cause of death in under one year of age in developing countries. Even in our century, its etiology is not clear, and there is no biomarker that is discriminative enough to predict the risk of suffering from it. Therefore, in this work, taking a public dataset on the lipidomic profile of babies who died from this syndrome compared to a control group, a univariate analysis was performed using the Mann–Whitney U test, with the aim of identifying the characteristics that enable discriminating between both groups. Those characteristics with a p-value less than or equal to 0.05 were taken; once these characteristics were obtained, classification models were implemented (random forests (RF), logistic regression (LR), support vector machine (SVM) and naive Bayes (NB)). We used seventy percent of the data for model training, subjecting it to a cross-validation (k = 5) and later submitting to validation in a blind test with 30% of the remaining data, which allows simulating the scenario in real life—that is, with an unknown population for the model. The model with the best performance was RF, since in the blind test, it obtained an AUC of 0.9, specificity of 1, and sensitivity of 0.8. The proposed model provides the basis for the construction of a SIDS risk prediction computer tool, which will contribute to prevention, and proposes lines of research to deal with this pathology.

Classification of Alzheimer's Disease Based on Deep Learning of Brain Structural and Metabolic Data

To improve the diagnosis and classification of Alzheimer's disease (AD), a modeling method is proposed based on the combining magnetic resonance images (MRI) brain structural data with metabolite levels of the frontal and parietal regions. First, multi-atlas brain segmentation technology based on T1-weighted images and edited magnetic resonance spectroscopy (MRS) were used to extract data of 279 brain regions and levels of 12 metabolites from regions of interest (ROIs) in the frontal and parietal regions. The t-test combined with false discovery rate (FDR) correction was used to reduce the dimensionality in the data, and MRI structural data of 54 brain regions and levels of 4 metabolites that obviously correlated with AD were screened out. Lastly, the stacked auto-encoder neural network (SAE) was used to classify AD and healthy controls (HCs), which judged the effect of classification method by fivefold cross validation. The results indicated that the mean accuracy of the five experimental model increased from 96 to 100%, the AUC value increased from 0.97 to 1, specificity increased from 90 to 100%, and F1 value increased from 0.97 to 1. Comparing the effect of each metabolite on model performance revealed that the gamma-aminobutyric acid (GABA) + levels in the parietal region resulted in the most significant improvement in model performance, with the accuracy rate increasing from 96 to 98%, the AUC value increased from 0.97 to 0.99 and the specificity increasing from 90 to 95%. Moreover, the GABA + levels in the parietal region was significantly correlated with Mini Mental State Examination (MMSE) scores of patients with AD (r = 0.627), and the F statistics were largest (F = 25.538), which supports the hypothesis that dysfunctional GABAergic system play an important role in the pathogenesis of AD. Overall, our findings support that a comprehensive method that combines MRI structural and metabolic data of brain regions can improve model classification efficiency of AD.

Mass Spectrometry-Based Analysis of Lipid Involvement in Alzheimer’s Disease Pathology—A Review

Irregularities in lipid metabolism have been linked to numerous neurodegenerative diseases. The roles of abnormal brain, plasma, and cerebrospinal fluid (CSF) lipid levels in Alzheimer’s disease (AD) onset and progression specifically have been described to a great extent in the literature. Apparent hallmarks of AD include, but are not limited to, genetic predisposition involving the APOE Ɛ4 allele, oxidative stress, and inflammation. A common culprit tied to many of these hallmarks is disruption in brain lipid homeostasis. Therefore, it is important to understand the roles of lipids, under normal and abnormal conditions, in each process. Lipid influences in processes such as inflammation and blood–brain barrier (BBB) disturbance have been primarily studied via biochemical-based methods. There is a need, however, for studies focused on uncovering the relationship between lipid irregularities and AD by molecular-based quantitative analysis in transgenic animal models and human samples alike. In this review, mass spectrometry as it has been used as an analytical tool to address the convoluted relationships mentioned above is discussed. Additionally, molecular-based mass spectrometry strategies that should be used going forward to further relate structure and function relationships of lipid irregularities and hallmark AD pathology are outlined.

A Nuclear Magnetic Resonance Spectroscopy Method in Characterization of Blood Metabolomics for Alzheimer's Disease

There is currently a crucial need for improved diagnostic techniques and targeted treatment methods for Alzheimer's disease (AD), a disease which impacts millions of elderly individuals each year. Metabolomic analysis has been proposed as a potential methodology to better investigate and understand the progression of this disease. In this report, we present our AD metabolomics results measured with high resolution magic angle spinning (HRMAS) nuclear magnetic resonance (NMR) on human blood plasma samples obtained from AD and non-AD subjects. Our study centers on developments of AD and non-AD metabolomics differentiating models with procedures of quality assurance (QA) and quality control (QC) through pooled samples. Our findings suggest that analysis of blood plasma samples using HRMAS NMR has the potential to differentiate between diseased and healthy subjects, which has important clinical implications for future improvements in AD diagnosis methodologies.

Random Forest Model in the Diagnosis of Dementia Patients with Normal Mini-Mental State Examination Scores

Background: Mini-Mental State Examination (MMSE) is the most widely used tool in cognitive screening. Some individuals with normal MMSE scores have extensive cognitive impairment. Systematic neuropsychological assessment should be performed in these patients. This study aimed to optimize the systematic neuropsychological test battery (NTB) by machine learning and develop new classification models for distinguishing mild cognitive impairment (MCI) and dementia among individuals with MMSE ≥ 26. Methods: 375 participants with MMSE ≥ 26 were assigned a diagnosis of cognitively unimpaired (CU) (n = 67), MCI (n = 174), or dementia (n = 134). We compared the performance of five machine learning algorithms, including logistic regression, decision tree, SVM, XGBoost, and random forest (RF), in identifying MCI and dementia. Results: RF performed best in identifying MCI and dementia. Six neuropsychological subtests with high-importance features were selected to form a simplified NTB, and the test time was cut in half. The AUC of the RF model was 0.89 for distinguishing MCI from CU, and 0.84 for distinguishing dementia from nondementia. Conclusions: This simplified cognitive assessment model can be useful for the diagnosis of MCI and dementia in patients with normal MMSE. It not only optimizes the content of cognitive evaluation, but also improves diagnosis and reduces missed diagnosis.

A deep learning framework with an embedded-based feature selection approach for the early detection of the Alzheimer's disease

Ageing is associated with various ailments including Alzheimer 's disease (AD), which is a progressive form of dementia. AD symptoms develop over a period of years and, unfortunately, there is no cure. Existing AD treatments can only slow down the progression of symptoms and thus it is critical to diagnose the disease at an early stage. To help improve the early diagnosis of AD, a deep learning-based classification model with an embedded feature selection approach was used to classify AD patients. An AD DNA methylation data set (64 records with 34 cases and 34 controls) from the GEO omnibus database was used for the analysis. Before selecting the relevant features, the data were preprocessed by performing quality control, normalization and downstream analysis. As the number of associated CpG sites was huge, four embedded-based feature selection models were compared and the best method was used for the proposed classification model. An Enhanced Deep Recurrent Neural Network (EDRNN) was implemented and compared to other existing classification models, including a Convolutional Neural Network (CNN), a Recurrent Neural Network (RNN), and a Deep Recurrent Neural Network (DRNN). The results showed a significant improvement in the classification accuracy of the proposed model as compared to the other methods.

Alterations in Glycerolipid and Fatty Acid Metabolic Pathways in Alzheimer's Disease Identified by Urinary Metabolic Profiling: A Pilot Study

An easily accessible and non-invasive biomarker for the early detection of Alzheimer's disease (AD) is needed. Evidence suggests that metabolic dysfunction underlies the pathophysiology of AD. While urine is a non-invasively collectable biofluid and a good source for metabolomics analysis, it is not yet widely used for this purpose. This small-scale pilot study aimed to examine whether the metabolic profile of urine from AD patients reflects the metabolic dysfunction reported to underlie AD pathology, and to identify metabolites that could distinguish AD patients from cognitively healthy controls. Spot urine of 18 AD patients (AD group) and 18 age- and sex-matched, cognitively normal controls (control group) were analyzed by mass spectrometry (MS). Capillary electrophoresis time-of-flight MS and liquid chromatography–Fourier transform MS were used to cover a larger range of molecules with ionic as well as lipid characteristics. A total of 304 ionic molecules and 81 lipid compounds of 12 lipid classes were identified. Of these, 26 molecules showed significantly different relative concentrations between the AD and control groups (Wilcoxon's rank-sum test). Moreover, orthogonal partial least-squares discriminant analysis revealed significant discrimination between the two groups. Pathway searches using the KEGG database, and pathway enrichment and topology analysis using Metaboanalyst software, suggested alterations in molecules relevant to pathways of glycerolipid and glycerophospholipid metabolism, thermogenesis, and caffeine metabolism in AD patients. Further studies of urinary metabolites will contribute to the early detection of AD and understanding of its pathogenesis.