Predicting clinical progression trajectories of early Alzheimer's disease patients.

BACKGROUND: Models for forecasting individual clinical progression trajectories in early Alzheimer's disease (AD) are needed for optimizing clinical studies and patient monitoring. METHODS: Prediction models were constructed using a clinical trial training cohort (TC; n = 934) via a gradient boosting algorithm and then evaluated in two validation cohorts (VC 1, n = 235; VC 2, n = 421). Model inputs included baseline clinical features (cognitive function assessments, APOE ε4 status, and demographics) and brain magnetic resonance imaging (MRI) measures. RESULTS: The model using clinical features achieved R2 of 0.21 and 0.31 for predicting 2-year cognitive decline in VC 1 and VC 2, respectively. Adding MRI features improved the R2 to 0.29 in VC 1, which employed the same preprocessing pipeline as the TC. Utilizing these model-based predictions for clinical trial enrichment reduced the required sample size by 20% to 49%. DISCUSSION: Our validated prediction models enable baseline prediction of clinical progression trajectories in early AD, benefiting clinical trial enrichment and various applications.

Plasma pTau217 predicts continuous brain amyloid levels in preclinical and early Alzheimer's disease.

BACKGROUND: This study investigated the potential of phosphorylated plasma Tau217 ratio (pTau217R) and plasma amyloid beta (Aß) 42/Aß40 in predicting brain amyloid levels measured by positron emission tomography (PET) Centiloid (CL) for Alzheimer's disease (AD) staging and screening. METHODS: Quantification of plasma pTau217R and Aß42/Aß40 employed immunoprecipitation-mass spectrometry. CL prediction models were developed on a cohort of 904 cognitively unimpaired, preclinical and early AD subjects and validated on two independent cohorts. RESULTS: Models integrating pTau217R outperformed Aß42/Aß40 alone, predicting amyloid levels up to 89.1 CL. High area under the receiver operating characteristic curve (AUROC) values (89.3% to 94.7%) were observed across a broad CL range (15 to 90). Utilizing pTau217R-based models for low amyloid levels reduced PET scans by 70.5% to 78.6%. DISCUSSION: pTau217R effectively predicts brain amyloid levels, surpassing cerebrospinal fluid Aß42/Aß40's range. Combining it with plasma Aß42/Aß40 enhances sensitivity for low amyloid detection, reducing unnecessary PET scans and expanding clinical utility. HIGHLIGHTS: Phosphorylated plasma Tau217 ratio (pTau217R) effectively predicts amyloid-PET Centiloid (CL) across a broad spectrum. Integrating pTau217R with Aß42/Aß40 extends the CL prediction upper limit to 89.1 CL. Combined model predicts amyloid status with high accuracy, especially in cognitively unimpaired individuals. This model identifies subjects above or below various CL thresholds with high accuracy. pTau217R-based models significantly reduce PET scans by up to 78.6% for screening out individuals with no/low amyloid.

Predictive signature development based on maximizing the area between receiver operating characteristic curves.

Development of marker signatures to predict treatment benefits for a new therapeutic is an important scientific component in advancing the drug discovery and is an important first step toward the goal of precision medicine. In this article, we focus on developing an algorithm to search for optimal linear combination of markers that maximizes the area between two receiver operating characteristic curves of the new therapeutic and the control groups without assuming any parametric model. We further generalize the proposed algorithm for predictive signature development to maximize the difference of Harrel's C-index of the new therapeutic and the control groups when the outcome of interest is time-to-event. The performance of this proposed method is evaluated and compared to existing methods via simulations and real clinical trial data.

VGF in Cerebrospinal Fluid Combined With Conventional Biomarkers Enhances Prediction of Conversion From MCI to AD.

BACKGROUND: Previous work has suggested that the brain and cerebrospinal fluid (CSF) levels of a neural protein involved in synaptic transmission, VGF (a noninitialism), may be altered in mild cognitive impairment (MCI) and Alzheimer Disease (AD). The objective of the current work is to examine the potential of CSF levels of a peptide derived from VGF to predict conversion from MCI to AD. MATERIALS AND METHODS: Using multivariate analytical approaches, the performance of the conventional biomarkers (CSF Aß1-42 and phosphorylated tau +/- hippocampal volume) was compared with the same biomarkers combined with CSF VGF peptide levels in a large publicly available data set from human subjects. RESULTS: It was observed that VGF peptides are lowered in CSF of patients with AD compared with controls and that combinations of CSF Aß1-42 and phosphorylated tau, hippocampal volume, and VGF peptide levels outperformed conventional biomarkers alone (hazard ratio=2.2 vs. 3.9), for predicting MCI to AD conversion. CONCLUSIONS: CSF VGF enhances the ability of conventional biomarkers to predict MCI to AD conversion. Future work will be needed to determine the specificity of VGF for AD versus other neurodegenerative diseases.

Intelligent and effective informatic deconvolution of "Big Data" and its future impact on the quantitative nature of neurodegenerative disease therapy.

Biomedical data sets are becoming increasingly larger and a plethora of high-dimensionality data sets ("Big Data") are now freely accessible for neurodegenerative diseases, such as Alzheimer's disease. It is thus important that new informatic analysis platforms are developed that allow the organization and interrogation of Big Data resources into a rational and actionable mechanism for advanced therapeutic development. This will entail the generation of systems and tools that allow the cross-platform correlation between data sets of distinct types, for example, transcriptomic, proteomic, and metabolomic. Here, we provide a comprehensive overview of the latest strategies, including latent semantic analytics, topological data investigation, and deep learning techniques that will drive the future development of diagnostic and therapeutic applications for Alzheimer's disease. We contend that diverse informatic "Big Data" platforms should be synergistically designed with more advanced chemical/drug and cellular/tissue-based phenotypic analytical predictive models to assist in either de novo drug design or effective drug repurposing.

Patient subgroup identification for clinical drug development.

Causal mechanism of relationship between the clinical outcome (efficacy or safety endpoints) and putative biomarkers, clinical baseline, and related predictors is usually unknown and must be deduced empirically from experimental data. Such relationships enable the development of tailored therapeutics and implementation of a precision medicine strategy in clinical trials to help stratify patients in terms of disease progression, clinical response, treatment differentiation, and so on. These relationships often require complex modeling to develop the prognostic and predictive signatures. For the purpose of easier interpretation and implementation in clinical practice, defining a multivariate biomarker signature in terms of thresholds (cutoffs/cut points) on individual biomarkers is preferable. In this paper, we propose some methods for developing such signatures in the context of continuous, binary and time-to-event endpoints. Results from simulations and case study illustration are also provided. Copyright © 2017 John Wiley & Sons, Ltd.

Big data to smart data in Alzheimer's disease: The brain health modeling initiative to foster actionable knowledge.

Massive investment and technological advances in the collection of extensive and longitudinal information on thousands of Alzheimer patients results in large amounts of data. These "big-data" databases can potentially advance CNS research and drug development. However, although necessary, they are not sufficient, and we posit that they must be matched with analytical methods that go beyond retrospective data-driven associations with various clinical phenotypes. Although these empirically derived associations can generate novel and useful hypotheses, they need to be organically integrated in a quantitative understanding of the pathology that can be actionable for drug discovery and development. We argue that mechanism-based modeling and simulation approaches, where existing domain knowledge is formally integrated using complexity science and quantitative systems pharmacology can be combined with data-driven analytics to generate predictive actionable knowledge for drug discovery programs, target validation, and optimization of clinical development.

A PRIM approach to predictive-signature development for patient stratification.

Patients often respond differently to a treatment because of individual heterogeneity. Failures of clinical trials can be substantially reduced if, prior to an investigational treatment, patients are stratified into responders and nonresponders based on biological or demographic characteristics. These characteristics are captured by a predictive signature. In this paper, we propose a procedure to search for predictive signatures based on the approach of patient rule induction method. Specifically, we discuss selection of a proper objective function for the search, present its algorithm, and describe a resampling scheme that can enhance search performance. Through simulations, we characterize conditions under which the procedure works well. To demonstrate practical uses of the procedure, we apply it to two real-world data sets. We also compare the results with those obtained from a recent regression-based approach, Adaptive Index Models, and discuss their respective advantages. In this study, we focus on oncology applications with survival responses.

Proceedings of the 14th European immunogenicity platform open symposium on immunogenicity of biopharmaceuticals.

Biologics have revolutionized disease management in many therapeutic areas by addressing unmet medical needs and overcoming resistance to standard-of-care treatment in numerous patients. However, the development of unwanted immune responses directed against these drugs, humoral and/or cellular, can hinder their efficacy and have safety consequences with various degrees of severity. Health authorities ask that a thorough immunogenicity risk assessment be conducted during drug development to incorporate an appropriate monitoring and mitigation plan in clinical studies. With the rapid diversification and complexification of biologics, which today include modalities such as multi-domain antibodies, cell-based products, AAV delivery vectors, and nucleic acids, developers are faced with the challenge of establishing a risk assessment strategy sometimes in the absence of specific regulatory guidelines. The European Immunogenicity Platform (EIP) Open Symposium on Immunogenicity of Biopharmaceuticals and its one-day training course gives experts and newcomers across academia, industry, and regulatory agencies an opportunity to share experience and knowledge to overcome these challenges. Here, we report the discussions that took place at the EIP's 14th Symposium, held in April 2023. The topics covered included immunogenicity monitoring and clinical relevance, non-clinical immunogenicity risk assessment, regulatory aspects of immunogenicity assessment and reporting, and the challenges associated with new modalities, which were discussed in a dedicated session.

Evaluation of plasma proteomic data for Alzheimer disease state classification and for the prediction of progression from mild cognitive impairment to Alzheimer disease.

Previous studies that have examined the potential for plasma markers to serve as biomarkers for Alzheimer disease (AD) have studied single analytes and focused on the amyloid-ß and τ isoforms and have failed to yield conclusive results. In this study, we performed a multivariate analysis of 146 plasma analytes (the Human DiscoveryMAP v 1.0 from Rules-Based Medicine) in 527 subjects with AD, mild cognitive impairment (MCI), or cognitively normal elderly subjects from the Alzheimer's Disease Neuroimaging Initiative database. We identified 4 different proteomic signatures, each using 5 to 14 analytes, that differentiate AD from control patients with sensitivity and specificity ranging from 74% to 85%. Five analytes were common to all 4 signatures: apolipoprotein A-II, apolipoprotein E, serum glutamic oxaloacetic transaminase, α-1-microglobulin, and brain natriuretic peptide. None of the signatures adequately predicted progression from MCI to AD over a 12- and 24-month period. A new panel of analytes, optimized to predict MCI to AD conversion, was able to provide 55% to 60% predictive accuracy. These data suggest that a simple panel of plasma analytes may provide an adjunctive tool to differentiate AD from controls, may provide mechanistic insights to the etiology of AD, but cannot adequately predict MCI to AD conversion.

Improving the Robustness of Variable Selection and Predictive Performance of Regularized Generalized Linear Models and Cox Proportional Hazard Models.

High-dimensional data applications often entail the use of various statistical and machine-learning algorithms to identify an optimal signature based on biomarkers and other patient characteristics that predicts the desired clinical outcome in biomedical research. Both the composition and predictive performance of such biomarker signatures are critical in various biomedical research applications. In the presence of a large number of features, however, a conventional regression analysis approach fails to yield a good prediction model. A widely used remedy is to introduce regularization in fitting the relevant regression model. In particular, a L1 penalty on the regression coefficients is extremely useful, and very efficient numerical algorithms have been developed for fitting such models with different types of responses. This L1-based regularization tends to generate a parsimonious prediction model with promising prediction performance, i.e., feature selection is achieved along with construction of the prediction model. The variable selection, and hence the composition of the signature, as well as the prediction performance of the model depend on the choice of the penalty parameter used in the L1 regularization. The penalty parameter is often chosen by K-fold cross-validation. However, such an algorithm tends to be unstable and may yield very different choices of the penalty parameter across multiple runs on the same dataset. In addition, the predictive performance estimates from the internal cross-validation procedure in this algorithm tend to be inflated. In this paper, we propose a Monte Carlo approach to improve the robustness of regularization parameter selection, along with an additional cross-validation wrapper for objectively evaluating the predictive performance of the final model. We demonstrate the improvements via simulations and illustrate the application via a real dataset.

CSF peptides from VGF and other markers enhance prediction of MCI to AD progression using the ATN framework.

The amyloid beta, tau, neurodegenerative markers framework has been proposed to serve as a system to classify and combine biomarkers for Alzheimer's Disease (AD). Although cerebrospinal (CSF) fluid AT (amyloid beta and tau)-based biomarkers have a well-established track record to distinguish AD from control subjects and to predict conversion from mild cognitive impairment (MCI) to AD, there is not an established non-tau based neurodegenerative ("N") marker from CSF. Here, we examine the ability of several candidate peptides in the CSF to serve as "N" markers to both classify disease state and predict MCI to AD conversion. We observed that although many putative N markers involved in synaptic processing and neuroinflammation were able to, when examined in isolation, distinguish MCI converters from non-converters, a derivative from VGF, when combined with AT markers, most strongly enhanced prediction of MCI to AD conversion. Low CSF VGF levels were also predictive of MCI to dementia conversion in the setting of normal AT markers, suggesting that it may serve as a very early predictor of dementia conversion. Other markers derived from neuronal pentraxin 2, GAP-43 and a 14-3-3 protein were also able to enhance MCI to AD prediction when used as a marker of neurodegeneration, but VGF had the highest predictive capacity. Thus, we propose that low levels of VGF in CSF may serve as "N" in the amyloid beta, tau, neurodegenerative markers framework to enhance the prediction of MCI to AD conversion.

The age-specific comorbidity burden of mild cognitive impairment: a US claims database study.

BACKGROUND: Identifying individuals with mild cognitive impairment (MCI) who are likely to progress to Alzheimer's disease and related dementia disorders (ADRD) would facilitate the development of individualized prevention plans. We investigated the association between MCI and comorbidities of ADRD. We examined the predictive potential of these comorbidities for MCI risk determination using a machine learning algorithm. METHODS: Using a retrospective matched case-control design, 5185 MCI and 15,555 non-MCI individuals aged ≥50 years were identified from MarketScan databases. Predictive models included ADRD comorbidities, age, and sex. RESULTS: Associations between 25 ADRD comorbidities and MCI were significant but weakened with increasing age groups. The odds ratios (MCI vs non-MCI) in 50-64, 65-79, and ≥ 80 years, respectively, for depression (4.4, 3.1, 2.9) and stroke/transient ischemic attack (6.4, 3.0, 2.1). The predictive potential decreased with older age groups, with ROC-AUCs 0.75, 0.70, and 0.66 respectively. Certain comorbidities were age-specific predictors. CONCLUSIONS: The comorbidity burden of MCI relative to non-MCI is age-dependent. A model based on comorbidities alone predicted an MCI diagnosis with reasonable accuracy.

Cerebrospinal fluid cytokine dynamics differ between Alzheimer disease patients and elderly controls.

The time courses of levels of multiple plasma and cerebrospinal fluid (CSF) cytokines in patients with Alzheimer disease (AD) and in age-matched control subjects were compared. Interleukin (IL)-1ß, IL-2, IL-6, IL-8, IL-10, IL-12p70, granulocyte-macrophage colony-stimulating factor, interferon-γ, and tumor necrosis factor alpha levels were measured 7 times over a 24-hour period in plasma and CSF using a lumbar catheter. Baseline plasma and CSF cytokine levels were found to be similar in AD and control subjects. However, the CSF levels of all measured cytokines, except IL-6 and IL-8, diverged over time between AD and control subjects, such that CSF cytokine levels in AD subjects were higher than in controls. This difference was greatest at 24 hours after the insertion of the lumbar catheter. In contrast, no differences in cytokine trajectories were seen in plasma. These data suggest that the neuroinflammatory response to lumbar catheter placement differs between AD and control subjects.

Comparison of Titer and Signal to Noise (S/N) for Determination of Anti-drug Antibody Magnitude Using Clinical Data from an Industry Consortium.

During biotherapeutic drug development, immunogenicity is evaluated by measuring anti-drug antibodies (ADAs). The presence and magnitude of ADA responses is assessed using a multi-tier workflow where samples are screened, confirmed, and titered. Recent reports suggest that the assay signal to noise ratio (S/N) obtained during the screening tier correlates well with titer. To determine whether S/N could more broadly replace titer, anonymized ADA data from a consortium of sponsors was collected and analyzed. Datasets from clinical programs with therapeutics of varying immunogenicity risk levels (low to high), common ADA assay platforms (ELISA and MSD) and formats (bridging, direct, solid-phase extraction with acid dissociation), and titration approaches (endpoint and interpolated) were included in the analysis. A statistically significant correlation between S/N and titer was observed in all datasets, with a strong correlation (Spearman's r > 0.8) in 11 out of 15 assays (73%). For assays with available data, conclusions regarding ADA impact on pharmacokinetics and pharmacodynamics were similar using S/N or titer. Subject ADA kinetic profiles were also comparable using the two measurements. Determination of antibody boosting in patients with pre-existing responses could be accomplished using similar approaches for titer and S/N. Investigation of factors that impacted the accuracy of ADA magnitude measurements revealed advantages and disadvantages to both approaches. In general, S/N had superior precision and ability to detect potentially low affinity/avidity responses compared to titer. This analysis indicates that S/N could serve as an equivalent and in some cases preferable alternative to titer for assessing ADA magnitude and evaluation of impact on clinical responses.

A hidden Markov model-based algorithm for identifying tumour subtype using array CGH data.

BACKGROUND: The recent advancement in array CGH (aCGH) research has significantly improved tumor identification using DNA copy number data. A number of unsupervised learning methods have been proposed for clustering aCGH samples. Two of the major challenges for developing aCGH sample clustering are the high spatial correlation between aCGH markers and the low computing efficiency. A mixture hidden Markov model based algorithm was developed to address these two challenges. RESULTS: The hidden Markov model (HMM) was used to model the spatial correlation between aCGH markers. A fast clustering algorithm was implemented and real data analysis on glioma aCGH data has shown that it converges to the optimal cluster rapidly and the computation time is proportional to the sample size. Simulation results showed that this HMM based clustering (HMMC) method has a substantially lower error rate than NMF clustering. The HMMC results for glioma data were significantly associated with clinical outcomes. CONCLUSIONS: We have developed a fast clustering algorithm to identify tumor subtypes based on DNA copy number aberrations. The performance of the proposed HMMC method has been evaluated using both simulated and real aCGH data. The software for HMMC in both R and C++ is available in ND INBRE website http://ndinbre.org/programs/bioinformatics.php.

Derivation of a new ADAS-cog composite using tree-based multivariate analysis: prediction of conversion from mild cognitive impairment to Alzheimer disease.

Model-based statistical approaches were used to compare the ability of the Alzheimer's Disease Assessment Scale-cognitive subscale (ADAS-cog), cerebrospinal fluid (CSF), fluorodeoxyglucose positron emission tomography and volumetric magnetic resonance imaging (MRI) markers to predict 12-month progression from mild cognitive impairment (MCI) to Alzheimer disease (AD). Using the Alzheimer's Disease Neuroimaging Initiative (ADNI) data set, properties of the 11-item ADAS-cog (ADAS.11), the 13-item ADAS-cog (ADAS.All) and novel composite scores were compared, using weighting schemes derived from the Random Forests (RF) tree-based multivariate model. Weighting subscores using the RF model of ADAS.All enhanced discrimination between elderly controls, MCI and AD patients. The ability of the RF-weighted ADAS-cog composite and individual scores, along with neuroimaging or biochemical biomarkers to predict MCI to AD conversion over 12 months was also assessed. Although originally optimized to discriminate across diagnostic categories, the ADAS. All, weighted according to the RF model, did nearly as well or better than individual or composite baseline neuroimaging or CSF biomarkers in prediction of 12-month conversion from MCI to AD. These suggest that a modified subscore weighting scheme applied to the 13-item ADAS-cog is comparable to imaging or CSF markers in prediction of conversion from MCI to AD at 12 months.

Serum Phosphatidylethanolamine and Lysophosphatidylethanolamine Levels Differentiate Alzheimer's Disease from Controls and Predict Progression from Mild Cognitive Impairment.

BACKGROUND: There is intense interest in the development of blood-based biomarkers, not only that can differentiate Alzheimer's disease (AD) from controls, but that can also predict conversion from mild cognitive impairment (MCI) to AD. Serum biomarkers carry the potential advantage over imaging or spinal fluid markers both in terms of cost and invasiveness. OBJECTIVE: Our objective was to measure the potential for serum lipid markers to differentiate AD from age-matched healthy controls as well as to predict conversion from MCI to AD. METHODS: Using a publicly-available dataset, we examined the relationship between baseline serum levels of 349 known lipids from 16 classes of lipids to differentiate disease state as well as to predict the conversion from MCI to AD. RESULTS: We observed that several classes of lipids (cholesteroyl ester, phosphatidylethanolamine, lysophosphatidylethanolamine, and acylcarnitine) differentiated AD from normal controls. Among these, only two classes, phosphatidylethanolamine (PE) and lysophosphatidylethanolamine (lyso-PE), predicted time to conversion from MCI to AD. Low levels of PE and high levels of lyso-PE result in two-fold faster median time to progression from MCI to AD, with hazard ratios 0.62 and 1.34, respectively. CONCLUSION: These data suggest that serum PE and lyso-PE may be useful biomarkers for predicting MCI to AD conversion. In addition, since PE is converted to lyso-PE by phospholipase A2, an important inflammatory mediator that is dysregulated in AD, these data suggest that the disrupted serum lipid profile here may be related to an abnormal inflammatory response early in the AD pathologic cascade.

Reported Hearing Loss in Alzheimer's Disease Is Associated With Loss of Brainstem and Cerebellar Volume.

Multiple epidemiological studies have revealed an association between presbycusis and Alzheimer's Disease (AD). Unfortunately, the neurobiological underpinnings of this relationship are not clear. It is possible that the two disorders share a common, as yet unidentified, risk factor, or that hearing loss may independently accelerate AD pathology. Here, we examined the relationship between reported hearing loss and brain volumes in normal, mild cognitive impairment (MCI) and AD subjects using a publicly available database. We found that among subjects with AD, individuals that reported hearing loss had smaller brainstem and cerebellar volumes in both hemispheres than individuals without hearing loss. In addition, we found that these brain volumes diminish in size more rapidly among normal subjects with reported hearing loss and that there was a significant interaction between cognitive diagnosis and the relationship between reported hearing loss and these brain volumes. These data suggest that hearing loss is linked to brainstem and cerebellar pathology, but only in the context of the pathological state of AD. We hypothesize that the presence of AD-related pathology in both the brainstem and cerebellum creates vulnerabilities in these brain regions to auditory deafferentation-related atrophy. These data have implications for our understanding of the potential neural substrates for interactions between hearing loss and AD.

Identification of distinct plasma biomarker signatures in patients with rapid and slow declining forms of COPD.

Chronic obstructive pulmonary disease (COPD) is a prevalent pulmonary disease characterized by a progressive decline in lung function. The identification of biomarkers capable of predicting the rate of lung function decline or capable of giving an early read on drug efficacy in clinical trials would be very useful. The aim of this study was to identify plasma biomarkers capable of accurately distinguishing patients with COPD from healthy controls. Eighty-nine plasma markers in 40 COPD patients and 20 healthy smoker controls were analyzed. The COPD patients were divided into two subgroups, rapid and slow decliners based on their rate of lung function decline measured over 15 years. Univariate analysis revealed that 25 plasma markers were statistically different between rapid decliners and controls, 4 markers were different between slow decliners and controls, and 10 markers were different between rapid and slow decliners (p < 0.05). Multivariate analysis led to the identification of groups of plasma markers capable of distinguishing rapid decliners from controls (signature 1), slow decliners from controls (signature 2) and rapid from slow decliners (signature 3) with over 90% classification accuracy. Importantly, signature 1 was shown to be longitudinally stable using plasma samples taken a year later from a subset of patients. This study describes a novel set of plasma markers differentiating slow from rapid decline of lung function in COPD. If validated in distinct and larger cohorts, the signatures identified will have important implications in both disease diagnosis, as well as the clinical evaluation of new therapies.