Transition. When our epilepsy genetics program was formally approved in 2019, my former division chief asked me what I was most excited about. I spontaneously replied that we finally got a fellowship and training program. This year marks the fifth anniversary of our epilepsy genetics fellowship, and I am stepping down in my role as the inaugural fellowship director. I wanted to mark this transition by pointing out five principles for teaching epilepsy genetics that have become relevant in the last few years. Spoiler alert: they are not what you think they are.
Epilepsy Genetics Graduation 2025. Our ENGIN fellowship team 2025 including (left to right from myself) our incoming fellow Riley Kessler, graduating fellow Laura McGarry, and current fellowship director Jill McKee. The CHOP Neurology graduation typically takes place in the Philadelphia College of Physicians aka the “hall of old white men with beards”. This is us under the portrait of Alfred Stillé (1813–1900) who advanced American medicine through education reforms and who help shape modern medical professionalism by being a critical part in founding the American Medical Association.
Teaching epilepsy genetics. We have come a long way. From constantly needing to explain why genetic testing in childhood epilepsy matters to starting the first precision medicine trials. Along the way, we have developed an entire body of knowledge that helps navigate the complex interplay of dynamic phenotypes and genetic findings that is increasingly formalized into official variant curation criteria. The role of a formal fellowship is to build a curriculum that ensures that this knowledge and experience can be conveyed to trainees within 12 months. Our fellows gain this knowledge and experience through weekly teaching sessions, dedicated trainee clinics, exposure to the patient population of one of the largest epilepsy genetic programs in the country, and interactions with epilepsy surgery, community epilepsy and child neurology programs. After three months, our fellows understand the basic principles. After six months, they feel comfortable across the various disease entities and decision-making frameworks. After nine months, fellows become independent, and we can step back and help fine-tune. This is not just theory, we have applied this framework now for five consecutive years.
Evolving concepts. The field is evolving rapidly and the epilepsy genetics I learned is unlike what we need to teach trainees today. Accordingly, the fives rules that I am covering in this blog post are not the basic principles that we want all epilepsy genetics provider to know. Yes, it is important that we know about historical origins of concepts such as GEFS+, the long road to discovering PRRT2, and the critical insights that Doose had to distinguish Myoclonic Astatic Epilepsy (MAE) from Lennox-Gastaut Syndrome (LGS). But again, the field is evolving and there are new principles that have emerged over the last few years. Here is what we want trainees to know.
Rule 1 – Know the rules. As an epilepsy genetics community, we need to pay close attention to the rules that make a variant pathogenic and appear on a genetic testing report in the first place. In my annual teaching session at AES, I have a dramatic slide that shows two comets on a collision course – our clinical concepts within pediatric epilepsy, including the various iterations of the ILAE classification and the concepts used by laboratory medicine and diagnostic labs. There is a growing discrepancy between both field and we need to be aware of this. Variant curation is formalized within the ACMG/AMP criteria and no matter what we think and feel about a variant, we need to understand what drives the various parameters that add to a variant’s pathogenicity, uncertainty, or benign-ness. ACMG/AMP criteria feel rigid and hard to adjust when we have a new tool, algorithm, or concept that seems to represent a major step forward. However, it is a formal diagnostic framework that has all the required safety checks. And these criteria are not immutable – for example, we have recently revised the formal ACMG/AMP criteria for epilepsy-related sodium channels. However, it is not the type of work that gives you a eureka moment or high-impact publication – it is a slowly advancing effort with lots of back and forth. Understanding this overall framework is critically important for anyone who commits to learning epilepsy genetics.
Rule 2 – Be aware of your own biopoetry. We all have our biases. Clinical expertise, research interests, favorite methods, topics that we have been excited about for several years or even decades. These stories often come together in unusual ways when we try to provide some background or justification for how a certain gene or variants connects to a particular clinical presentation. It reminds you of a mouse model you studied long ago, a patient you saw as a trainee, or a gene group that one of your colleagues from graduate school has worked on for the last decade. This “groundswell” of knowledge can be extremely helpful to maintain momentum and create distinction between the large number of variants and phenotypes that we try to put into context. However, over the last few years, I have also encountered the flipside of this. With the best of intentions, we tend to spin stories and let our imagination run wild. We create “biopoetry” about genes and variants, which may create obstacles at some point. Your favorite gene may not have a gene-disease relationship, the seemingly suspicious location of a variant may not shift classification after all. We don’t need to dispose of biopoetry altogether, we just need to realize that it represents a bias in our thinking.
Rule 3 – There is a correct place for phenotypes and function. Here are some of my personal biases that I am still struggling with. As a clinician, it is sometimes seemingly obvious to me that a specific phenotype virtually must be due to a particular gene. Likewise, we typically consider functional assessments such as voltage-clamp electrophysiological recording for ion channel variants as the ground truth. However, within the broader framework of trying to understand if a variant is causative, we often have the burden of proof backwards. In brief, the burden is on us to provide evidence rather than simply accepting that phenotypes and function data are sacrosanct. We often use the framework of an “uphill battle” – every variant is uncertain until proven otherwise. Yes, there are formal frameworks for phenotypes and functional data to add to variant pathogenicity. However, they add some evidence but often do not tilt the scale in favor of pathogenicity. To contribute more and to drive and dictate variant pathogenicity, it would take extraordinarily strong evidence that we have not accumulated yet. In brief, we simply do not know the predictive value of hemiclonic seizures for Dravet Syndrome in the broader population and we do not know the frequency of SCN1A population variants that decrease current density by up to 20%. Yes, rare disorders are rare and such information may be difficult to obtain. However, the fact that variant assessment is increasingly formalized into objective rules is a benefit to the epilepsy genetics field. We will need this kind of certainty as the foundation for precision medicine therapies.
Rule 4 – Accept variant uncertainty and minimize explanatoriness. A few years ago, I introduced the concept of “explanatoriness”, trying to create a framework for how clinical decision represents a decision layer above formal variant interpretation. However, I have since come to realize that this framework may be imprecise. There is a gap between what we see as clinically obvious and what variant criteria can tell us. For example, I would still maintain that a transmembrane non-population SCN1A variant in child with the clinical picture of Dravet Syndrome is causative, even if the de novo status of this variant cannot be confirmed. However, I would like to rephrase the framework behind this. It is not a conceptually different layer of decision-making, but the fact that the formal curation criteria are not there yet. In an ideal world, the decision-making framework for calling this variant pathogenic would be transparent and evidence-based and not due to clinical intuition. I would also like to put careful parentheses around the concept of explanatoriness to avoid one major pitfall in epilepsy genetics: embracing uncertainty. Even though we try to avoid uncertainly in clinical decision making at all costs, there are variants that simply remain unclear in their contribution to an individual’s phenotype. Rather than invoking concepts such as clinical intuition too rapidly, our goal is to teach trainees how to communicate uncertainty.
Rule 5 – Embrace the active stance and therapeutic microdecisions. Genetic testing either leads to an immediate clinical decision (such as starting/stopping a medication) or only provides closure and helps with family planning? This might be the impression you get from the literature at times. Let me take a step back and examine this narrative a bit more closely. In brief, this is the view of genetic testing in the epilepsies looking backwards, a retrospective chart review or retrospective family interview. However, this is not what actually happens in real-time clinical care. Think of a genetic diagnosis as an overall framework that allows us to compare an individual’s trajectory to what would be expected. And we actually make dozens if not hundreds of decisions along the way that are informed by the genetic diagnosis. Most importantly features that cannot be explained by an individual’s diagnosis and raise red flags. I refer to the many (and often) subtle clinical considerations as “therapeutic microdecisions”. This concept represents an active stance as opposed to the black-or-white framework of immediate actionability in rare disease. And it also shifts the focus on how to demonstrate the benefits of making a genetic diagnosis, from retrospective analysis of individual features to embracing a comprehensive analysis of real-world clinical data.
