Epilepsy biomarkers - Toward etiology and pathology specificity

Abstract

A biomarker is a characteristic that is measured as an indicator of normal biologic processes, pathogenic processes, or responses to an exposure or intervention, including therapeutic interventions. Biomarker modalities include molecular, histologic, radiographic, or physiologic characteristics. In 2015, the FDA-NIH Joint Leadership Council developed the BEST Resource (Biomarkers, EndpointS, and other Tools) to improve the understanding and use of biomarker terminology in biomedical research, clinical practice, and medical product development. The BEST biomarker categories include: (a) susceptibility/risk biomarkers, (b) diagnostic biomarkers, (c) monitoring biomarkers, (d) prognostic biomarkers, (e) predictive biomarkers, (f) pharmacodynamic/response biomarkers, and (g) safety biomarkers. Here we review 30 epilepsy biomarker studies that have identified (a) diagnostic biomarkers for epilepsy, epileptogenesis, epileptogenicity, drug-refractoriness, and status epilepticus - some of the epileptogenesis and epileptogenicity biomarkers can also be considered prognostic biomarkers for the development of epilepsy in subjects with a given brain insult, (b) predictive biomarkers for epilepsy surgery outcome, and (c) a response biomarker for therapy outcome. The biomarker modalities include plasma/serum/exosomal and cerebrospinal fluid molecular biomarkers, brain tissue molecular biomarkers, imaging biomarkers, electrophysiologic biomarkers, and behavioral/cognitive biomarkers. Both single and combinatory biomarkers have been described. Most of the reviewed biomarkers have an area under the curve >0.800 in receiver operating characteristics analysis, suggesting high sensitivity and specificity. As discussed in this review, we are in the early phase of the learning curve in epilepsy biomarker discovery. Many of the seven biomarker categories lack epilepsy-related biomarkers. There is a need for epilepsy biomarker discovery using proper, statistically powered study designs with validation cohorts, and the development and use of novel analytical methods. A strategic roadmap to discuss the research priorities in epilepsy biomarker discovery, regulatory issues, and optimization of the use of resources, similar to those devised in the cancer and Alzheimer's disease research areas, is also needed.

Keywords: Antiepileptogenesis; Area under the curve; Diagnosis; Electroencephalogram; Epileptogenesis; Magnetic resonance imaging; Receiver operating characteristics; microRNA.

Fig. 1.

Epileptogenesis is an evolving target - implications for biomarker discovery. (A) In a follow-up cohort, the proportion of subjects with epilepsy will increase from time-point t₁ to t₂. Thus, selection of the analysis endpoint (based on previous follow-up studies in a given animal model or patient cohort) significantly affects the power calculations related to the number of subjects needed to conduct a sufficiently-powered epileptogenesis biomarker study. Also, the ability of the biomarker to detect epileptogenesis is affected by the time-point selected. (B) At the cellular level, neuronal excitability will increase during epileptogenesis, and will occasionally fluctuate over the seizure threshold. The diagnostic biomarker of epileptogenesis (i.e., prognostic biomarker for development of epilepsy after an epileptogenic brain insult) should be sensitive and specific to differentiate subjects with seizure susceptibility fluctuating over the threshold from those in whom it stays below the threshold. Predictive bio-markers of therapy response should be able to identify a subject in which the treatment (Rx) prevents an increase in seizure susceptibility over the threshold or repairs the tissue to bring the seizure susceptibility below the threshold.

Fig. 2.

Epileptogenic process and need of biomarkers for different purposes in basic science, clinical practice, and medical product (including pharmacotherapy) development. Epileptogenesis is initiated by an “epilepsy gene” or various types of acute brain insults or chronic neurodegenerative diseases. Some of the conditions can present with status epilepticus. The entire epileptogenic process is modulated by an individual’s genetic background, microbiota, and exposome (non-genetic exposures of an individual in a lifetime, e.g., lifestyle, medications, etc.; Miller and Jones, 2014; Wild, 2012). Epileptogenesis continues after epilepsy diagnosis (i.e., occurrence of the first unprovoked seizure). Different phases of epileptogenic process can benefit from different biomarker types: susceptibility/risk biomarkers (BM), diagnostic biomarkers, prognostic biomarkers, pharmacodynamic/response biomarkers, predictive biomarkers, monitoring biomarkers, and safety biomarkers. As epilepsy can occur with a myriad of co-morbidities, epilepsy biomarkers can also apply to co-morbidities as well as to progression to remission/cure or occurrence of sudden unexpected death (SUDEP).

Fig. 3.

Tissue ecosystem during epileptogenesis as a biomarker source. (A) Secondary tissue damage and temporal regulation of underlying molecular networks will progress and vary with the development of epilepsy (epileptogenesis). (B) Mechanistic bio-markers (BM) representing the regulated gene networks and molecular pathways originate in different cellular sources, including (b1) apoptotic and necrotic cells, (b2) activated astrocytes, (b3) activate microglia, (b4) intraparenchymal T cells, (b5) activated pericytes, (b6) vascular remodeling, including vascular damage and blood-brain-barrier leakage, and consequent angiogenesis, (b7) myelin damage, (b8) axonal sprouting (e.g., mossy fiber sprouting in the dentate gyrus), (b9) iron deposits indicating hemorrhage, and (b10) calcifications (e.g., in the thalamus after lateral fluid-percussion injury). (C) A coronal Nissl sections showing a schematic of bio-marker expression at possible injury sites (perilesional cortex, thalamus, hippocampus [HC]) after lateral fluid-percussion injury in the rat brain. Various biomarkers are expressed (red and blue dots), some of which can be secreted into the blood (red dots) or cerebrospinal fluid (CSF; blue dots). The efficiency and region dependency of biomarker secretion is unknown. Also, the possible exposure of a given biomarker to degradation during the secretion phase is poorly understood. Some cellular changes can be imaged, for example, with magnetic resonance imaging or spectroscopy. Some of the changes can be measured with electrophysiology.

Fig. 4.

A schematics of the biomarker discovery process to identify diagnostic biomarkers for epileptogenesis. (A) Targetselection and lead identification. Biomarker discovery can be hypothesis or non-hypothesis driven (e.g., tissue molecular omics, imaging). A small number of epileptogenic and non-epileptogenic animals (~10) is analyzed to find differences in markers and generate a biomarker (BM) candidate list. In case of omics findings, confirmation of positive hits is established using lower-throughput independent analysis platforms. The top leads (typically < 10) will be taken for further analysis. (B) Statistically powered pre-clinical validation studies demonstrate that the candidate biomarker differentiates epileptogenic from non-epileptogenic animals with high sensitivity and specificity. The area under the curve (AUC) in receiver operating characteristics (ROC) analysis should be > 0.800. In parallel, assay platforms should be developed to achieve a highly sensitive and specific assay for high-throughput analysis. Its analytical performance, including accuracy, precision, linearity, limit of detection, and limit of quantification should be determined. Also, standardization and quality control (QC) in tissue sampling (e.g., plasma quality) should be performed both for preclinical and clinical samples. (C) Clinical validation in a target patient population should estimate the frequency of true positives and false positives (ROC analysis). It should analyze co-variates, including age, sex, ethnicity, injury, nutrition, sports and training, other medical conditions and their treatment. Studies can be retrospective on data available from repositories or prospective. After regulatory approval (D) product launch and marketing is accompanied by generation of best practice guidelines that define indications for the use of a given biomarker, and educational programs to train and educate clinicians and patients for their use, benefits, and limitations. It should be noted that the experimental design of the discovery process for identifying susceptibility/risk biomarkers, diagnostic biomarkers, prognostic biomarkers, monitoring biomarkers, predictive biomarkers, pharmacodynamic/response biomarkers, or safety biomarkers needs to be tailored accordingly.

Fig. 5.

Statistical issues related to bio-marker analysis. (A) Density plots showing the proportions of TN (true negative), TP (true positive), FN (false negative), and FP (false positive) cases in two hypothetical diagnostic tests [test A and test B; simulated data, 200 cases per group] that were used to diagnose epileptogenesis after brain injury. The black dashed line indicates a cut-off test value (e.g., plasma miRNA concentration) that is used to calculate the receiver operating characteristics (ROC) curve (see below). Movement of the cut-off line to the left or right increases sensitivity or specificity, respectively. Sensitivity and specificity are calculated from the distribution of positive (epileptogenic) and negative (non- epileptogenic) cases: sensitivity = TP / (TP + FN), specificity = TN / (TN + FP). (B) Sensitivity and specificity of test A and test B are plotted across a series of cut-off test values as ROC curves, revealing the area under the curve (AUC) values for each test. In this example, the ROC curve of test A is better than that of test B in discriminating the non-epileptogenic from epileptogenic subjects (p < 1.23e-14, bootstrap test). (C) Comparison of the two ROC curves with similar AUC values. Partial AUC (pAUC) can be calculated for both the sensitivity and specificity. In our example, the pAUC is calculated and standardized for 90%−100% sensitivity and 90%−100% specificity. ROC curve A performs better that B when high biomarker specificity is needed. ROC curve B is superior to A when high sensitivity is needed. (D) Illustration of a standardized pAUC and the relevant areas: randomAUC = area(ABCD), pAUC = area (AEFD), perfectAUC = (AGHD). All data were generated and statistics performed with R version 3.4.2 and RStudio version 1.1.383 (https://www.R-project.org; https://www.rstudio.com/) using R packages pROC (Robin et al., 2011) and ggplot2 (Wickham, 2009).

Fig. 6.

Current status of circulating epilepsy biomarker discovery - what do the biomarkers monitor? (A) Bioinformatics analysis of targets for diagnostic microRNA (miRNA) biomarkers in the circulation and cerebrospinal fluid (CSF) listed in Table 3. For analysis, we collected all predicted targets for each miRNA from TargetScanHuman (Release 7.1: June 2016). Then, we generated gene lists and submitted the lists to the gene-annotation enrichment analysis using the DAVID Functional Annotation Tool (DAVID Bioinformatics Resources 6.8). In all panels, the x-axis shows the number of predicted targets and the y-axis shows the 10 most enriched gene-annotations for each miRNA in the DAVID analysis. Despite variability in the tissue (serum, plasma, CSF) sampled for miRNA biomarker analysis in different patient populations, variable sampling times relative to the epileptogenic process, and variable miRNA analysis platforms, we found many miRNAs bio-markers to target similar molecular functions, including alternative splicing, phosphoprotein, coiled coil as well as nucleic and cytoplasm functions. (B) A circos plot representing all differentially expressed circulating miRNAs and protein biomarkers found in epilepsy studies. Note that only few miRNAs and proteins were found to be dysregulated in multiple studies (panel B; miR-106b and miR-146a upregulated, red line; miR-194 downregulated, blue line). miR-301a was upregulated in one study and downregulated in another (green line, Table 3). Currently, there are no circulating biomarker studies involving animal models, except the one studying HMGB1 (Walker et al., 2017). No same biomarkers were found when plasma/serum was compared with plasma exosomes or CSF. Interestingly, our bioinformatics prediction of targets for dysregulated miRNAs revealed that miR-130a and miR-301a share the same targets (yellow lines in panel B). In addition, striking similarities were found between gene annotation enrichment of the targets of miR-146a and miR-4521 (purple line in panel B; note that the number of predicted targets for these two miRNAs was different). We suggest that when biomarker studies are compared with each other, bioinformatics prediction of the miRNA targets and molecular pathways to which they belong could be one approach to investigate mechanisms behind the increased/decreased miRNA content in body fluids. Color codes: red text, upregulated; blue text, downregulated; red line, upregulated in multiple studies; blue line, downregulated in multiple studies; green line; opposite expression pattern. Abbreviations: CSF, cerebrospinal fluid; miRNA, microRNA.