Biomarker selection for medical diagnosis using the partial area under the ROC curve

Current State of Knowledge on Blood and Tissue-Based Biomarkers for Opisthorchis viverrini-induced Cholangiocarcinoma: A Review of Prognostic, Predictive, and Diagnostic Markers

Cholangiocarcinoma (CCA) is a prevalent cancer in Southeast Asia, with Opisthorchis viverrini (O.viverrini) infection being the primary risk factor. Most CCA cases in this region are diagnosed at advanced stages, leading to unfavorable prognoses. The development of stage-specific biomarkers for Opisthorchis viverrini-induced cholangiocarcinoma (Ov-CCA) holds crucial significance, as it facilitates early detection and timely administration of curative interventions, effectively mitigating the high morbidity and mortality rates associated with this disease in the Great Mekong region. Biomarkers are a promising approach for early detection, prognosis, and targeted treatment of CCA. Disease-specific biomarkers facilitate early detection and enable monitoring of therapy effectiveness, allowing for any necessary corrections. This review provides an overview of the potential O. viverrini-specific molecular biomarkers and important markers for diagnosing and monitoring Ov-CCA, discussing their prognostic, predictive, and diagnostic value. Despite the limited research in this domain, several potential biomarkers have been identified, encompassing both worm-induced and host-induced factors. This review offers a thorough examination of historical and contemporary progress in identifying biomarkers through multiomics techniques, along with their potential implications for early detection and treatment.

Plasma IgG aggregates as biomarkers for multiple sclerosis

We recently reported that multiple sclerosis (MS) plasma contains IgG aggregates and induces complement-dependent neuronal cytotoxicity (Zhou et al., 2023). Using ELISA, we report herein that plasma IgG levels in the aggregates can be used as biomarkers for MS. We enriched the IgG aggregates from samples of two cohorts (190 MS and 160 controls) by collecting flow-through after plasma binding to Protein A followed by detection of IgG subclass. We show that there are significantly higher levels of IgG1, IgG3, and total IgG antibodies in MS IgG aggregates, with an AUC >90%; higher levels of IgG1 distinguish secondary progressive MS from relapsing-remitting MS (AUC = 91%). Significantly, we provided the biological rationale for MS plasma IgG biomarkers by demonstrating the strong correlation between IgG antibodies and IgG aggregate-induced neuronal cytotoxicity. These non-invasive, simple IgG-based blood ELISA assays can be adapted into clinical practice for diagnosing MS and SPMS and monitoring treatment responses.

Metabolic Fingerprints of Effective Fluoxetine Treatment in the Prefrontal Cortex of Chronically Socially Isolated Rats: Marker Candidates and Predictive Metabolites

The increasing prevalence of depression requires more effective therapy and the understanding of antidepressants' mode of action. We carried out untargeted metabolomics of the prefrontal cortex of rats exposed to chronic social isolation (CSIS), a rat model of depression, and/or fluoxetine treatment using liquid chromatography-high resolution mass spectrometry. The behavioral phenotype was assessed by the forced swim test. To analyze the metabolomics data, we employed univariate and multivariate analysis and biomarker capacity assessment using the receiver operating characteristic (ROC) curve. We also identified the most predictive biomarkers using a support vector machine with linear kernel (SVM-LK). Upregulated myo-inositol following CSIS may represent a potential marker of depressive phenotype. Effective fluoxetine treatment reversed depressive-like behavior and increased sedoheptulose 7-phosphate, hypotaurine, and acetyl-L-carnitine contents, which were identified as marker candidates for fluoxetine efficacy. ROC analysis revealed 4 significant marker candidates for CSIS group discrimination, and 10 for fluoxetine efficacy. SVM-LK with accuracies of 61.50% or 93.30% identified a panel of 7 or 25 predictive metabolites for depressive-like behavior or fluoxetine effectiveness, respectively. Overall, metabolic fingerprints combined with the ROC curve and SVM-LK may represent a new approach to identifying marker candidates or predictive metabolites for ongoing disease or disease risk and treatment outcome.

Gene Expression Profiling as a Novel Diagnostic Tool for Neurodegenerative Disorders

There is a lack of effective diagnostic biomarkers for neurodegenerative disorders (NDDs). Here, we established gene expression profiles for diagnosing Alzheimer’s disease (AD), Parkinson’s disease (PD), and vascular (VaD)/mixed dementia. Patients with AD had decreased APOE, PSEN1, and ABCA7 mRNA expression. Subjects with VaD/mixed dementia had 98% higher PICALM mRNA levels, but 75% lower ABCA7 mRNA expression than healthy individuals. Patients with PD and PD-related disorders showed increased SNCA mRNA levels. There were no differences in mRNA expression for OPRK1, NTRK2, and LRRK2 between healthy subjects and NDD patients. APOE mRNA expression had high diagnostic accuracy for AD, and moderate accuracy for PD and VaD/mixed dementia. PSEN1 mRNA expression showed promising accuracy for AD. PICALM mRNA expression was less accurate as a biomarker for AD. ABCA7 and SNCA mRNA expression showed high-to-excellent diagnostic accuracy for AD and PD, and moderate-to-high accuracy for VaD/mixed dementia. The APOE E4 allele reduced APOE expression in patients with different APOE genotypes. There was no association between PSEN1, PICALM, ABCA7, and SNCA gene polymorphisms and expression. Our study suggests that gene expression analysis has diagnostic value for NDDs and provides a liquid biopsy alternative to current diagnostic methods.

Identification of a miRNA Panel with a Potential Determinant Role in Patients Suffering from Periodontitis

In recent years, the role of microRNA (miRNA) in post-transcriptional gene regulation has advanced and supports strong evidence related to their important role in the regulation of a wide range of fundamental biological processes. Our study focuses on identifying specific alterations of miRNA patterns in periodontitis compared with healthy subjects. In the present study, we mapped the major miRNAs altered in patients with periodontitis (n = 3) compared with healthy subjects (n = 5), using microarray technology followed by a validation step by qRT-PCR and Ingenuity Pathways Analysis. Compared to healthy subjects, 159 differentially expressed miRNAs were identified among periodontitis patients, of which 89 were downregulated, and 70 were upregulated, considering a fold change of ±1.5 as the cut-off value and p ≤ 0.05. Key angiogenic miRNAs (miR-191-3p, miR-221-3p, miR-224-5p, miR-1228-3p) were further validated on a separate cohort of patients with periodontitis versus healthy controls by qRT-PCR, confirming the microarray data. Our findings indicate a periodontitis-specific miRNA expression pattern representing an essential issue for testing new potential diagnostic or prognostic biomarkers for periodontal disease. The identified miRNA profile in periodontal gingival tissue was linked to angiogenesis, with an important molecular mechanism that orchestrates cell fate.

A circulating microRNA panel as a novel dynamic monitor for oral squamous cell carcinoma

Oral squamous cell carcinoma (OSCC) has high recurrence and mortality rates despite advances in diagnosis and treatment. Therefore, it is necessary to identify new biomarkers for early detection, efficient monitoring, and prognosis prediction. Since microRNA (miRNA) is stable and detectable in serum, it has been reported to inform the diagnosis and monitor disease progression through liquid biopsy. In this study, a circulating specific miRNA panel in OSCC patients was developed, and its usefulness as a dynamic monitor was validated. Small RNAs were extracted from the serum of OSCC patients (n = 4) and normal controls (n = 6) and profiled using next-generation sequencing. NGS identified 42 differentially expressed miRNAs (DEmiRNAs) in serum between patients with OSCC and healthy controls, with threefold differences (p < 0.05). Combining the 42 DEmiRNAs and The Cancer Genome Atlas (TCGA) databases OSCC cohort, 9 overlapping DEmiRNAs were screened out. Finally, 4 significantly up-regulated miRNAs (miR-92a-3p, miR-92b-3p, miR-320c and miR-629-5p) were identified from OSCC patients via validation in the Chungnam National University Hospital cohort. Application of the specific miRNA panel for distinguishing OSCC patients from healthy controls produced specificity and sensitivity of 97.8 and 74%, respectively. In addition, the serum levels of these 4 miRNAs significantly decreased after complete surgical resection and increased after recurrence. We suggest that circulating 4-miRNA panel might be promising non-invasive predictors for diagnosing and monitoring the progression of patients with OSCC.

Immunogenomic Biomarkers and Validation in Lynch Syndrome

Lynch syndrome (LS) is an inherited disorder in which affected individuals have a significantly higher-than-average risk of developing colorectal and non-colorectal cancers, often before the age of 50 years. In LS, mutations in DNA repair genes lead to a dysfunctional post-replication repair system. As a result, the unrepaired errors in coding regions of the genome produce novel proteins, called neoantigens. Neoantigens are recognised by the immune system as foreign and trigger an immune response. Due to the invasive nature of cancer screening tests, universal cancer screening guidelines unique for LS (primarily colonoscopy) are poorly adhered to by LS variant heterozygotes (LSVH). Currently, it is unclear whether immunogenomic components produced as a result of neoantigen formation can be used as novel biomarkers in LS. We hypothesise that: (i) LSVH produce measurable and dynamic immunogenomic components in blood, and (ii) these quantifiable immunogenomic components correlate with cancer onset and stage. Here, we discuss the feasibility to: (a) identify personalised novel immunogenomic biomarkers and (b) validate these biomarkers in various clinical scenarios in LSVH.

Unveiling Chemical Interactions Between Plants and Fungi Using Metabolomics Approaches

Metabolomics has been extensively used in clinical studies in the search for new biomarkers of human diseases. However, this approach has also been highlighted in agriculture and biological sciences, once metabolomics studies have been assisting researchers to deduce new chemical mechanisms involved in biological interactions that occur between microorganisms and plants. In this sense, the knowledge of the biological role of each metabolite (virulence factors, signaling compounds, antimicrobial metabolites, among others) and the affected biochemical pathways during the interaction contribute to a better understand of different ecological relationships established in nature. The current chapter addresses five different applications of the metabolomics approach in fungal-plant interactions research: (1) Discovery of biomarkers in pathogen-host interactions, (2) plant diseases diagnosis, (3) chemotaxonomy, (4) plant defense, and (5) plant resistance; using mass spectrometry and/or nuclear magnetic resonance spectroscopy, which are the techniques most used in metabolomics.

1H NMR metabolomics analysis of oil palm stem tissue infected by Ganoderma boninense based on field severity Indices

Basal stem rot disease (BSR) caused by G. boninense affects most oil palm plants in Southeast Asia. This disease can be fatal to palm oil production. BSR shows no signs on the tree in the early stages of infection. Therefore, it is essential to find an approach that can detect BSR disease in oil palm, especially at any level of disease severity in the field. This study aims to identify biomarkers of BSR disease in oil palm stem tissue based on various disease severity indices in the field using ¹H NMR-based metabolomics analysis. The crude extract of oil palm stem tissue with four disease severity indices was analyzed by ¹H NMR metabolomics. Approximately 90 metabolites from oil palm stem tissue were identified.Twenty of these were identified as metabolites that significantly differentiated the four disease severity indices. These metabolites include the organic acid group, the carbohydrate group, the organoheterocyclic compound group, and the benzoid group. In addition, different tentative biomarkers for different disease severity indices were also identified. These tentative biomarkers consist of groups of organic acids, carbohydrates, organoheterocyclic compounds, nitrogenous organic compounds, and benzene. There are five pathways in oil palm that are potentially affected by BSR disease.

Normalized sensitivity of multi-dimensional body composition biomarkers for risk change prediction

The limitations of BMI as a measure of adiposity and health risks have prompted the introduction of many alternative biomarkers. However, ranking diverse biomarkers from best to worse remains challenging. This study aimed to address this issue by introducing three new approaches: (1) a calculus-derived, normalized sensitivity score (NORSE) is used to compare the predictive power of diverse adiposity biomarkers; (2) multiple biomarkers are combined into multi-dimensional models, for increased sensitivity and risk discrimination; and (3) new visualizations are introduced that convey complex statistical trends in a compact and intuitive manner. Our approach was evaluated on 23 popular biomarkers and 6 common medical conditions using a large database (National Health and Nutrition Survey, NHANES, N ~ 100,000). Our analysis established novel findings: (1) regional composition biomarkers were more predictive of risk than global ones; (2) fat-derived biomarkers had stronger predictive power than weight-related ones; (3) waist and hip are always elements of the strongest risk predictors; (4) our new, multi-dimensional biomarker models yield higher sensitivity, personalization, and separation of the negative effects of fat from the positive effects of lean mass. Our approach provides a new way to evaluate adiposity biomarkers, brings forth new important clinical insights and sets a path for future biomarker research.

Machine learning-based biomarkers identification from toxicogenomics - Bridging to regulatory relevant phenotypic endpoints

One of the major challenges in realization and implementations of the Tox21 vision is the urgent need to establish quantitative link between in-vitro assay molecular endpoint and in-vivo regulatory-relevant phenotypic toxicity endpoint. Current toxicomics approach still mostly rely on large number of redundant markers without pre-selection or ranking, therefore, selection of relevant biomarkers with minimal redundancy would reduce the number of markers to be monitored and reduce the cost, time, and complexity of the toxicity screening and risk monitoring. Here, we demonstrated that, using time series toxicomics in-vitro assay along with machine learning-based feature selection (maximum relevance and minimum redundancy (MRMR)) and classification method (support vector machine (SVM)), an "optimal" number of biomarkers with minimum redundancy can be identified for prediction of phenotypic toxicity endpoints with good accuracy. We included two case studies for in-vivo carcinogenicity and Ames genotoxicity prediction, using 20 selected chemicals including model genotoxic chemicals and negative controls, respectively. The results suggested that, employing the adverse outcome pathway (AOP) concept, molecular endpoints based on a relatively small number of properly selected biomarker-ensemble involved in the conserved DNA-damage and repair pathways among eukaryotes, were able to predict both Ames genotoxicity endpoints and in-vivo carcinogenicity in rats. A prediction accuracy of 76% with AUC = 0.81 was achieved while predicting in-vivo carcinogenicity with the top-ranked five biomarkers. For Ames genotoxicity prediction, the top-ranked five biomarkers were able to achieve prediction accuracy of 70% with AUC = 0.75. However, the specific biomarkers identified as the top-ranked five biomarkers are different for the two different phenotypic genotoxicity assays. The top-ranked biomarkers for the in-vivo carcinogenicity prediction mainly focused on double strand break repair and DNA recombination, whereas the selected top-ranked biomarkers for Ames genotoxicity prediction are associated with base- and nucleotide-excision repair The method developed in this study will help to fill in the knowledge gap in phenotypic anchoring and predictive toxicology, and contribute to the progress in the implementation of tox 21 vision for environmental and health applications.

Identification of Potential Biomarkers for Anti-PD-1 Therapy in Melanoma by Weighted Correlation Network Analysis

Melanoma is the most malignant form of skin cancer, which seriously threatens human life and health. Anti-PD-1 immunotherapy has shown clinical benefits in improving patients' overall survival, but some melanoma patients failed to respond. Effective therapeutic biomarkers are vital to evaluate and optimize benefits from anti-PD-1 treatment. Although the establishment of immunotherapy biomarkers is well underway, studies that identify predictors by gene network-based approaches are lacking. Here, we retrieved the existing datasets (GSE91061, GSE78220 and GSE93157, 79 samples in total) on anti-PD-1 therapy to explore potential therapeutic biomarkers in melanoma using weighted correlation network analysis (WGCNA), function validation and clinical corroboration. As a result, 13 hub genes as critical nodes were traced from the key module associated with clinical features. After receiver operating characteristic (ROC) curve validation by an independent dataset (GSE78220), six hub genes with diagnostic significance were further recovered. Moreover, these six genes were revealed to be closely associated not only with the immune system regulation, immune infiltration, and validated immunotherapy biomarkers, but also with excellent prognostic value and significant expression level in melanoma. The random forest prediction model constructed using these six genes presented a great diagnostic ability for anti-PD-1 immunotherapy response. Taken together, IRF1, JAK2, CD8A, IRF8, STAT5B, and SELL may serve as predictive therapeutic biomarkers for melanoma and could facilitate future anti-PD-1 therapy.

Redefining the Practical Utility of Blood Transcriptome Biomarkers in Inflammatory Bowel Diseases

BACKGROUND AND AIMS

The study investigates the practical utility of whole-blood gene expression profiling to diagnose inflammatory bowel diseases [IBDs].

METHODS

The discovery cohorts included 102 and 51 paediatric IBD patients and controls, and 95 and 46 adult IBD patients and controls, respectively. The replication cohorts included 447 and 76 paediatric IBD patients and controls, and 271 and 108 adult IBD patients and controls, respectively. In the discovery phase, RNA samples extracted from whole peripheral blood were analysed using RNA-Seq, and the predictive values of selected biomarkers were validated using quantitative polymerase chain reaction [qPCR].

RESULTS

In all, 15 differentially expressed transcripts [adjusted p ≤0.05] were selected from the discovery sequencing datasets. The receiver operating characteristic curves and area under the curve [ROC-AUC] in replication analyses showed high discriminative power [AUC range, 0.91-0.98] for 11 mRNAs in paediatric patients with active IBD. By contrast, the AUC-ROC values ranged from 0.63 to 0.75 in comparison among inactive paediatric IBDs and active/inactive adult IBDs, indicating a lack of discriminative power. The best multi-mRNA diagnostic classifier showed moderate discriminative power [AUC = 0.81] for paediatric inactive IBD, but was not able to discriminate active or inactive adult IBD patients from controls. The AUC-ROC values did not confirm an ability of the mRNAs abundances to discriminate between active ulcerative colitis and active Crohn's disease in paediatric or adult populations.

CONCLUSIONS

This study identifies and validates blood transcriptional biomarkers that could be used in clinical settings as diagnostic predictors of IBD clinical activity in paediatric, but not adult, IBD patients.

Method to quantify cytokines and chemokines in mouse brain tissue using Bio-Plex multiplex immunoassays

This protocol describes how to prepare mouse brain tissue for quantification of multiple inflammatory mediators using a multiplex bead-based immunoassay. It is important to have methods that allow quantification of multiple analytes from small amounts of tissue. Bio-Plex is a Luminex xMAP-based multiplex bead-based immunoassay technology that permits simultaneous analysis of up to 100 analytes from a single tissue sample. This assay has been used extensively to investigate analytes in plasma and serum samples as well as cultured and primary cells. Here, we describe a method for simultaneous analysis of 33 different inflammatory cytokines and chemokines from mouse brain tissue using the Bio-Plex Pro Mouse Chemokine Panel 33-Plex.

Influence Analysis for the Area Under the Receiver Operating Characteristic Curve

Classification measures play essential roles in the assessment and construction of classifiers. Hence, determining how to prevent these measures from being affected by individual observations has become an important problem. In this paper, we propose several indexes based on the influence function and the concept of local influence to identify influential observations that affect the estimate of the area under the receiver operating characteristic curve (AUC), an important and commonly used measure. Cumulative lift charts are also used to equipoise the disagreements among the proposed indexes. Both the AUC indexes and the graphical tools only rely on the classification scores, and both are applicable to classifiers that can produce real-valued classification scores. A real data set is used for illustration.

Data aggregation at the level of molecular pathways improves stability of experimental transcriptomic and proteomic data

High throughput technologies opened a new era in biomedicine by enabling massive analysis of gene expression at both RNA and protein levels. Unfortunately, expression data obtained in different experiments are often poorly compatible, even for the same biologic samples. Here, using experimental and bioinformatic investigation of major experimental platforms, we show that aggregation of gene expression data at the level of molecular pathways helps to diminish cross- and intra-platform bias otherwise clearly seen at the level of individual genes. We created a mathematical model of cumulative suppression of data variation that predicts the ideal parameters and the optimal size of a molecular pathway. We compared the abilities to aggregate experimental molecular data for the 5 alternative methods, also evaluated by their capacity to retain meaningful features of biologic samples. The bioinformatic method OncoFinder showed optimal performance in both tests and should be very useful for future cross-platform data analyses.

Support Vector Algorithms for Optimizing the Partial Area under the ROC Curve

The area under the ROC curve (AUC) is a widely used performance measure in machine learning. Increasingly, however, in several applications, ranging from ranking to biometric screening to medicine, performance is measured not in terms of the full area under the ROC curve but in terms of the partial area under the ROC curve between two false-positive rates. In this letter, we develop support vector algorithms for directly optimizing the partial AUC between any two false-positive rates. Our methods are based on minimizing a suitable proxy or surrogate objective for the partial AUC error. In the case of the full AUC, one can readily construct and optimize convex surrogates by expressing the performance measure as a summation of pairwise terms. The partial AUC, on the other hand, does not admit such a simple decomposable structure, making it more challenging to design and optimize (tight) convex surrogates for this measure. Our approach builds on the structural SVM framework of Joachims ( 2005 ) to design convex surrogates for partial AUC and solves the resulting optimization problem using a cutting plane solver. Unlike the full AUC, where the combinatorial optimization needed in each iteration of the cutting plane solver can be decomposed and solved efficiently, the corresponding problem for the partial AUC is harder to decompose. One of our main contributions is a polynomial time algorithm for solving the combinatorial optimization problem associated with partial AUC. We also develop an approach for optimizing a tighter nonconvex hinge loss-based surrogate for the partial AUC using difference-of-convex programming. Our experiments on a variety of real-world and benchmark tasks confirm the efficacy of the proposed methods.

Performance of variable selection methods using stability-based selection

BACKGROUND

Variable selection is frequently carried out during the analysis of many types of high-dimensional data, including those in metabolomics. This study compared the predictive performance of four variable selection methods using stability-based selection, a new secondary selection method that is implemented in the R package BioMark. Two of these methods were evaluated using the more well-known false discovery rate (FDR) as well.

RESULTS

Simulation studies varied factors relevant to biological data studies, with results based on the median values of 200 partial area under the receiver operating characteristic curve. There was no single top performing method across all factor settings, but the student t test based on stability selection or with FDR adjustment and the variable importance in projection (VIP) scores from partial least squares regression models obtained using a stability-based approach tended to perform well in most settings. Similar results were found with a real spiked-in metabolomics dataset. Group sample size, group effect size, number of significant variables and correlation structure were the most important factors whereas the percentage of significant variables was the least important.

CONCLUSIONS

Researchers can improve prediction scores for their study data by choosing VIP scores based on stability variable selection over the other approaches when the number of variables is small to modest and by increasing the number of samples even moderately. When the number of variables is high and there is block correlation amongst the significant variables (i.e., true biomarkers), the FDR-adjusted student t test performed best. The R package BioMark is an easy-to-use open-source program for variable selection that had excellent performance characteristics for the purposes of this study.

DNA methylation status is more reliable than gene expression at detecting cancer in prostate biopsy

Background:

We analysed critically the potential usefulness of RNA- and DNA-based biomarkers in supporting conventional histological diagnostic tests for prostate carcinoma (PCa) detection.

Methods:

Microarray profiling of gene expression and DNA methylation was performed on 16 benign prostatic hyperplasia (BPH) and 32 cancerous and non-cancerous prostate samples extracted by radical prostatectomy. The predictive value of the selected biomarkers was validated by qPCR-based methods using tissue samples extracted from the 58 prostates and, separately, using 227 prostate core biopsies.

Results:

HOXC6, AMACR and PCA3 expression showed the best discrimination between PCa and BPH. All three genes were previously reported as the most promising mRNA-based markers for distinguishing cancerous lesions from benign prostate lesions; however, none were sufficiently sensitive and specific to meet the criteria for a PCa diagnostic biomarker. By contrast, DNA methylation levels of the APC, TACC2, RARB, DGKZ and HES5 promoter regions achieved high discriminating sensitivity and specificity, with area under the curve (AUCs) reaching 0.95−1.0. Only a small overlap was detected between the DNA methylation levels of PCa-positive and PCa-negative needle biopsies, with AUCs ranging between 0.854 and 0.899.

Conclusions:

DNA methylation-based biomarkers reflect the prostate malignancy and might be useful in supporting clinical decisions for suspected PCa following an initial negative prostate biopsy.