Epilepsy affects millions of people around the world. What inspired your team to investigate this genetic form of epilepsy, and why is understanding its causes so important?
Epilepsy is one of the most common neurological disorders, affecting nearly 50 million people worldwide. Seizure control has improved enormously, yet approximately one third of patients have drug-resistant epilepsy, a reality that can quietly reshape a person’s education, work, independence and safety. For decades, the diagnostic pathway has rested on three pillars: clinical history, electroencephalography (EEG), and magnetic resonance imaging (MRI). In many patients, MRI identifies the structural substrate, such as hippocampal sclerosis, focal cortical dysplasia, tumours, vascular malformations, or developmental anomalies, which guides medical and surgical decisions.
But one question kept returning to us in the reading room: what do we tell the patient whose seizures are severe and disabling, yet whose MRI looks entirely normal? Such patients were once labelled “MRI-negative” or “cryptogenic,” as though the brain held no answer. That assumption is what inspired our work. A normal scan, we came to believe, does not mean there is nothing wrong. It often means the cause lies below the resolution of our images, in the biology of the brain itself. Establishing the cause matters, because a named diagnosis carries prognostic information, permits counselling and family screening, and increasingly opens the door to treatment directed at the underlying mechanism.
For readers without a scientific background, could you explain how certain genes help the brain develop normally, and what happens when these genes do not function as they should?
Cortical development is an exquisitely choreographed process. Before birth, neurons are generated, migrate to their destined layers, and assemble into functional networks under tight genetic control. Some of these genes act as regulators of cellular growth and proliferation, instructing cells when to divide and, equally importantly, when to stop.
Our recent study, published in Neuroradiology, examined one such regulatory system: GATOR1-related epilepsies, caused by pathogenic variants in the DEPDC5, NPRL2 and NPRL3 genes. Together these genes encode the GATOR1 complex, a key negative regulator of the mTOR signalling pathway, which governs neuronal growth, migration and cortical organisation during development. When this molecular brake fails, mTOR signalling runs unchecked, resulting in malformations of cortical development. GATOR1 variants are now recognised as one of the commonest genetic causes of focal epilepsy, in which seizures arise from a circumscribed region of the brain.
Your research revealed that changes in these genes can lead to a wide variety of brain changes. What were the most important discoveries, and how do they improve our understanding of epilepsy?
The most striking observation from our study was the extraordinary heterogeneity of neuroimaging findings. While focal cortical dysplasia, a localised disorganisation of cortical architecture, remains the classical hallmark, our patients also demonstrated diffuse cortical malformations, generalised cerebral atrophy, and, importantly, completely normal MRI examinations despite harbouring pathogenic mutations.
This concept extends beyond GATOR1-related disorders. Many genetic epilepsies, including ion channelopathies, synaptopathies, chromatin remodelling disorders, and metabolic epilepsies, may present with normal or nonspecific neuroimaging findings. Conversely, identical genetic mutations can produce markedly different imaging phenotypes, even among members of the same family.
Brain MRI is an important tool in diagnosing neurological disorders. How does your study help doctors better recognize and interpret brain changes associated with genetic epilepsy?
The role of the neuroradiologist has expanded accordingly. Rather than searching only for an overt lesion, we now look for subtle imaging biomarkers that may point toward an underlying molecular diagnosis. Lesion detection has improved markedly with volumetric acquisitions at high spatial resolution, dedicated epilepsy protocols, susceptibility-weighted imaging, diffusion tensor imaging, quantitative volumetry, functional MRI and advanced postprocessing. Artificial intelligence and machine learning are further enhancing detection of subtle dysplasia that may escape even experienced observers. By cataloguing the imaging spectrum associated with a defined genetic pathway, studies such as ours give radiologists a framework for recognising when a scan, including a normal one, should prompt genetic evaluation.
Some people with epilepsy have normal-looking brain scans despite experiencing severe seizures. How does your research help explain this, and why is it an important finding?
This directly addresses a longstanding paradox: how a patient may experience frequent, disabling seizures while the scan appears reassuringly normal. The abnormality was never absent. It lay beyond the resolution of conventional imaging, residing in molecular and microstructural organisation rather than in macroscopic anatomy. Recognising this reframes MRI-negative epilepsy from a diagnostic dead end into a diagnosis awaiting the right tools, and it changes the counselling such patients receive.
How can combining genetic testing with advanced brain imaging improve diagnosis, treatment decisions, and personalized care for people living with epilepsy?
Comprehensive genetic testing is now embedded in modern epilepsy evaluation. Targeted gene panels rapidly screen hundreds of genes associated with epilepsy; whole exome sequencing has become a cornerstone in unexplained drug-resistant epilepsy; and whole genome sequencing extends detection to structural variants, noncoding mutations and repeat expansions. Sequencing at very high depth can identify somatic mosaicism, in which mutations are confined to a small population of brain cells, an important mechanism underlying focal cortical dysplasia and many epilepsies driven by mTOR dysregulation.
The greatest value lies not in the diagnostic label alone, but in how it alters management. A pathogenic variant carries prognostic weight, enables genetic counselling and family screening, and increasingly directs targeted therapy. Inhibition of the mTOR pathway in selected disorders is the clearest example of a shift from symptom suppression toward mechanism-based treatment. Genetics has likewise reshaped epilepsy surgery. Resection remains highly effective in carefully selected patients, yet seizure freedom is not guaranteed even after apparently complete lesionectomy, because the epileptogenic process may extend beyond the visible lesion as microscopic dysplasia, multifocal abnormality or widespread network alteration. Surgical planning is therefore best undertaken in a multidisciplinary setting integrating imaging, electrophysiology, genetics and neuropathology.
Looking ahead, what are the next major challenges in this field, and how do you hope your research will contribute to better diagnosis, treatment, or quality of life for people affected by epilepsy?
Significant challenges remain, and they are of three distinct kinds. A large proportion of the variants identified on sequencing are of uncertain significance, leaving families without a definitive answer. Genotype–phenotype correlation is still imperfect: we cannot reliably predict which imaging pattern a given variant will produce, nor explain why the same variant yields different appearances in different individuals. And access to advanced imaging and sequencing remains uneven, scarcest in precisely those settings that carry the greatest epilepsy burden.
The second of these is where the clearest opportunity lies. Radiogenomics, the systematic correlation of imaging phenotypes with genetic data, seeks to make that relationship predictable in both directions: inferring a likely molecular diagnosis from the scan, and anticipating the imaging appearance of a known variant. The other two will require different remedies, namely functional validation of uncertain variants and equitable access to technology.
Looking ahead, we expect the boundary between imaging and sequencing to keep blurring. Advances in sequencing technology, transcriptomics, spatial genomics, digital pathology and computational image analysis are expected to uncover mechanisms currently beyond recognition and to enable genuinely personalised therapeutic strategies. The goal is a future in which every patient, including those with an apparently normal scan, receives an accurate diagnosis and a therapy chosen for the biology of their own brain.
For any reader drawn to this field, the central lesson is a hopeful one. A normal MRI is no longer the end of the search; it is often the beginning. By learning to read every scan, even the normal ones, as a clue to the molecular architecture beneath, we are moving epilepsy care from managing seizures to understanding the mechanism of the disease and ultimately treating them.
