How brains learn diseases, and their cures

Kelly Clancy
Jun 13 · 6 min read

My father is a pediatric neurologist, with a speciality in epilepsy. He’s always loved the kids he worked with, and treated them like family. Whenever one of his patients died, he’d bring me and my sisters to their funeral. Some I’d known for years from tagging along on rounds, or having watched them grow up through the quanta of Christmas cards and back-to-school photos. I remember the funeral of a girl my age, who had been far more outgoing and lively than me, her now-dark face caked in yellow makeup. Though they’d never found a drug that could control her seizures, she’d seemed otherwise healthy and normal, and her death had come as a total surprise. This was my first brush with Sudden Unexpected Death in Epilepsy (SUDEP). SUDEP is the most common cause of death in patients with refractory seizures, and remains one of the most devastating mysteries in the field of epilepsy. There are currently no known biomarkers to identify patients at risk for SUDEP — hence the Sudden, hence the Unexpected.

Having come to neuroscience by way of physics, I confess I often forget a crucial point: neurons are cells, not just immutable contact points on a circuit board over which voltage plays in prescribed patterns. Not only is the circuit board layout shifting, always rewiring itself in learning, but activity sets off internal programs within cells, making them more or less active, or plastic, tuning their metabolism and protein production. Furthermore, at the gross level, the brain isn’t a black box sealed in the skull, but exerts its influence on every conceivable bodily process. Neurological disorders are rarely, if ever, confined to the brain. This may seem obvious, but it bears consideration. When brain areas are over or under-active, this can reverberate through other somatic systems, causing plastic changes in all manner of body parts: neurons love to learn. In fact, non-neuronal cells learn too: signals like oxygen or nutrition levels can alter a cell’s transcriptional landscape, resulting in tissue remodelling, for example.

Epilepsy is particularly interesting, as it’s a prime example of a learned disease. “Kindling” is a common idea in the literature: once someone has had one or more seizures, they’re much more likely to become epileptic, as the brain learns this problematic activity pattern. Some epileptic patients report that they’ve discovered certain thought patterns that help them control or minimize an oncoming seizure (these patients feel a pre-ictal ‘aura’ that clues them in), so we know there may be conscious strategies which certain patients can use to volitionally control their disease. But I’d argue that neural circuits may have evolved or learned other, less helpful ways of suppressing disease-related activity—in the case of uncontrollable epilepsy, through depriving the brain of oxygen, thereby quenching activity. I posit we might use this framing of brains and bodies ‘learning’ their own cures to better understand common comorbidities of neurological disorders.

SUDEP is a big clue in this. It’s increasingly evident that cardio-pulmonary issues underlie a majority of SUDEP cases. While most SUDEP events go unwitnessed, in a retrospective study of observed SUDEP cases caught while patients were in the hospital, researchers found that patient breathing ceased before heart failure in all 16 observed events (Rivlin et al., 2013). There is no doubt that SUDEP involves a complex network of effects, but the effect of repeated exposure to low oxygen levels (chronic intermittent hypoxia) is consistent with many of the known abnormalities of SUDEP patients (Giaccia, Simon, & Johnson, 2004). Hypoxia has a profound effect on tissues, from altering the physiology of certain ion channels in the heart and lungs to long-term changes in genetic transcription and vascular remodeling (Kemp & Peers, 2007; Ling et al., 2001; Nei, 2009; Richerson, 2010). Up to 40% of SUDEP cases present cardiac fibrosis in autopsy, which can be a result of chronic intermittent hypoxia (Ling et al., 2001; P-Codrea, et al., 2005), and damage was most commonly found in the heart structure most vulnerable to ischemic damage (the subendocardial myocardium, for the keen). SUDEP patients also tend to have had post-seizure cerebral depression, which is also strongly linked with hypoxia (Takano et al., 2007). All of this suggests that recurring low oxygen levels might be the cause of the myriad tissue and neural remodeling events that ultimately result in patient death.

The link between epilepsy and breathing problems is well documented, but not well understood. In some patients, epileptic activity may simply affect brain stem circuits controlling breathing. Every patient has a different locus of seizure activity — the part of the brain a seizure plays out on. But SUDEP doesn’t only happen in patients whose seizures originate in the part of the brain that controls breathing. There may be a more fundamental physiological link underlying seizures and breathing: blood acidification, as happens in hypoxic conditions, is anti-convulsive. Thus, the hypoxic response may not simply be a direct effect of a seizure on the brain stem, but an evolved mechanism, or one learned by neural circuits, to shut down epileptic activity during otherwise intractable seizures [1]. This then leads to a vicious cycle: breathing circuits are plastic throughout life, and strongly shaped by hypoxia (So, 2008), so even if a patient’s seizures never directly reach the brain stem, their chronic seizures could still cause a gradual degradation of the neural circuits controlling breathing through repeated hypoxia [2]. The brain, then, might be learning activity patterns that help control a particular disease state, which is adaptive in the short term but ultimately deleterious: a drastic stop-gap to manage otherwise intractable epilepsy.

The brain is sometimes too good at doing its job. Many neurological diseases involve some component of being ‘learned’ by neural circuits: chronic pain, neuropathy/neuralgia, and migraine all come to mind. When considering how to find predictive biomarkers for neurological conditions, or how to address their symptoms, it’s crucial to remember that neurological diseases don’t live in the brain alone. They can affect all body systems, and some of the most deleterious symptoms might result from the brain attempting to ‘medicate’ itself with whatever other systems it has control over. Studies with this framing in mind might help lead to the discovery of simple interventions, and further refine our understanding to identify functional and molecular biomarkers of medical mysteries such as SUDEP [3].


  1. Interestingly, SUDEP is more prevalent among males, and males tend to be more sensitive to hypoxia, suffering worse outcomes. This has been found true both in clinical studies and in studies involving multiple animal models (Hill & Fitch, 2012; Mayoral, Omar, & Penn, 2009; Mirza, 2015; Pérez-Crespo et al., 2005). This sexual dimorphism may be due, in part, to the gene G6PD, an X-linked gene related to oxidative stress, which has been implicated in hypoxia-induced damage (Chettimada et al., 2014; Gao, L., et al., 2004). It’s also worth noting that in a rodent model, sleep apnea-induced plasticity of the breathing circuit was normalized with antioxidant treatment, which may similarly show a protective effect in high SUDEP risk patients. Therefore, markers of oxidative stress may play an important role in detecting and preventing SUDEP.
  2. Some researchers believe that SUDEP is related to another medical mystery, Sudden Infant Death Syndrome (SIDS). One of the best known risk factors for SIDS is prenatal nicotine exposure, and nicotine exposure is known to alter receptor expression in oxygen-sensing neural circuits; therefore both diseases may result from plasticity of the brain areas responsible for controlling breathing.
  3. SUDEP has also been associated with a variety of genes implicated in cardiac arrhythmias, as well as several anti-convulsive drugs that also make patients prone to arrhythmias (Klassen et al., 2013; Glasscock, 2014; So, 2008). However, genes do not deterministically dictate SUDEP risk. Given that genetic tests are costly, and only suggestive of susceptibility, a functional measurement such as apnea would likely be a more effective biomarker.

The Spike

The science of the brain, from the scientists of the brain

Thanks to Mark Humphries

Kelly Clancy

Written by

Neuroscientist/writer, previously for Harper's, The New Yorker, Nautilus, and others.

The Spike

The Spike

The science of the brain, from the scientists of the brain

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade