Approximately 5 million new cases of epilepsy are diagnosed each year. Some have genetic causes, whereas many occur due to an earlier insult to the brain. Since epileptogenesis can take days, weeks, or months, this creates windows of opportunity for a treatment to disrupt or mitigate the process, preventing epilepsy altogether or reducing disease severity (Simonato et al., 2021). Clinical trials that used antiseizure drugs after insults such as stroke or trauma failed to prevent epilepsy (Trinka and Brigo, 2014; Chang et al., 2022). The need for specific antiepileptogenic drugs, designed to disrupt processes underlying epileptogenesis, is not the only challenge. Many people who experience a central nervous system (CNS) injury never develop epilepsy. Treating everyone regardless of risk is at a minimum extremely costly and likely risks harm. We will also need to know which drug to give for which epileptogenic insult and when. The development and application of biomarkers of epileptogenesis will help solve these problems. Minimally invasive ways to monitor the process is underway with sufficient sensitivity and specificity. A biomarker of epileptogenesis should be present before the first spontaneous seizure occurs and might be specific for a certain etiology or broad.

We increasingly understand the complex cell and molecular changes associated with a variety of epileptogenic processes and the timescales over which they operate (Engel and Pitkanen, 2020). The changes generated give rise to altered network neurophysiology and the circulation of cell components and molecules into local and distant fluid environments. Technologies capable of direct or indirect detection of these signatures of epileptogenesis exist. They include brain imaging, neurophysiological signals detected by electroencephalogram (EEG) analysis, and assays that measure circulating molecules in biofluids such as blood. There have been exciting and important advances in each of these approaches in recent years, and we have an increasing understanding of how they mechanistically link to disease. For example, immune cells continuously patrol and respond to molecular signals from the brain and contain transcriptional profiles that reflect brain disease states (Alves de Lima et al., 2020). Disruption of the blood–brain barrier, a common consequence of brain injury, as well as seizures, provides a physical route by which molecules or extracellular vesicles can transfer from the brain into the circulation. We also know that communication is bidirectional, with recent discoveries that circulating immune cells and gut bacteria release molecules that are taken up by and can direct brain cell function (Kur et al., 2020; Needham et al., 2022).

Section 4 of this book covers our current understanding as well as future directions in biomarkers of epileptogenesis as discovered in preclinical models and the ongoing efforts to validate these biomarkers in human studies. The section comprises five chapters. We begin with EEG biomarkers of epileptogenesis (Chapter 36 by Kamintsky et al.). EEG biomarkers are highly attractive as they comprise a standard tool for epilepsy diagnosis and have the advantage of reflecting underlying brain activity with high temporal, although not spatial, resolution. The latest preclinical and clinical research includes advances in signal processing that reflect network changes and may provide sufficient sensitivity and specificity to be useful to detect and monitor epileptogenesis.

Circulating molecules that can be detected in biofluids, including blood and cerebrospinal fluid, are attractive as biomarkers because their measurement can be simple and cost-effective. Emerging candidates include small noncoding RNAs called microRNAs. These appear stabile in biofluids, lend themselves to rapid quantitative measurement, and display tissue-specific expression (some are only expressed by specific cell types in the brain). Several brain-enriched and inflammation-associated proteins also show biomarker potential. There are limitations as well, since they lack the anatomic resolution of imaging and EEG biomarkers. Again, progress is assessed with a view to the studies needed for clinical validation and companion technology for their detection (see Chapter 37 by Henshall and Simonato).

Not all biomarkers require sophisticated technologies for detection. As brain networks change, so to do accompanying behaviors, a manifestation of altered cognitive, emotional, sensory, and executive functions. Shared pathomechanisms likely explain this relationship. Chapter 38 by Ali et al. focuses on these behavioral biomarkers, how they can be monitored, and their translational potential.

Brain imaging technologies provide another opportunity to identify biomarkers of epileptogenesis. This includes using functional magnetic resonance imaging (MRI), positron emission tomography (PET), and other modalities that detect disturbed brain function at various scales spanning from cellular to molecular with varying degrees of anatomic and temporal resolution. Chapter 39 by Koepp et al. looks closely at the latest advances as well as limitations of each modality for quantifying epileptogenic changes to structures and their molecular composition.

Finally, the discipline of computer science is increasingly providing methods to handle the large, complex, and heterogeneous data and to find within it signals that are missed. This includes machine learning and artificial intelligence approaches which researchers are now bringing together to search for signals across and within biomarker modalities (see Chapter 40 by Ciszek et al.).

Each chapter includes future-facing questions that must now be answered. These include the standardization, structured, and systematic approaches spanning preclinical discovery, through to planning for validation in clinical trials. Major hurdles remain around defining how much change and when represents an actionable time for therapeutic intervention. Ultimately, the sensitivity and specificity needed for a clinically actionable biomarker of epileptogenesis may be a multimodal risk score generated by an integration of neurophysiological, imaging, and circulating molecules.