Refractory epilepsy, also called drug-resistant epilepsy, is epilepsy that continues to cause seizures despite treatment with at least two appropriately chosen anti-seizure medications. About one in three people with epilepsy fall into this category. The formal definition, established by an international task force of the International League Against Epilepsy, requires that both medications were well-tolerated, used at adequate doses (alone or in combination), and still failed to achieve sustained seizure freedom.
Why Two Failed Medications Is the Threshold
The two-drug cutoff isn’t arbitrary. Data consistently show that if a first medication doesn’t control seizures, a second has a reasonable chance of working. But after two well-chosen medications have failed, the odds of a third, fourth, or fifth drug achieving full seizure control drop sharply, often to single-digit percentages. That inflection point is why the diagnosis triggers a shift in approach: rather than simply cycling through more drugs, the focus expands to surgical evaluation, devices, and dietary therapies.
Why Some Brains Resist Medication
No single explanation accounts for drug resistance, but two biological theories have the strongest evidence behind them.
The first is the transporter theory. The blood-brain barrier contains molecular pumps whose job is to keep foreign substances out of the brain. In some people with refractory epilepsy, these pumps are overexpressed, actively pushing medication back into the bloodstream before it can reach brain tissue in therapeutic concentrations. The most studied of these pumps is called P-glycoprotein, but several others play similar roles.
The second is the target theory. Anti-seizure medications work by binding to specific molecular targets in brain cells, most commonly sodium channels that regulate electrical firing. In drug-resistant epilepsy, these targets can undergo structural changes, either as a result of the disease itself or from repeated seizures over time. When the shape of the target shifts, the medication no longer fits as well and loses its ability to dampen abnormal electrical activity. Studies in both animal models and human brain tissue removed during epilepsy surgery have confirmed these changes, particularly a loss of the medication’s ability to regulate sodium channel behavior.
These mechanisms aren’t mutually exclusive. Some patients may have overactive drug pumps and altered targets simultaneously, compounding resistance.
The Evaluation Process
Once epilepsy is recognized as drug-resistant, the next step is a comprehensive evaluation to determine whether surgery or other targeted treatments are an option. This typically happens at a specialized epilepsy center and unfolds in phases.
The first phase includes three core tests: a high-resolution MRI of the brain, video-EEG monitoring (where you’re recorded on camera while your brain’s electrical activity is tracked continuously, sometimes for days), and a detailed neuropsychological assessment to map memory, language, and cognitive function. The MRI follows a specific epilepsy protocol with multiple sequences designed to detect subtle abnormalities like scarring in the hippocampus, malformations of brain development, small tumors, and vascular lesions. This specialized protocol can identify nearly all epilepsy-related brain lesions.
If those initial tests don’t clearly pinpoint the seizure source, additional imaging and electrical studies can be added. These include PET scans that reveal areas of reduced brain metabolism between seizures, SPECT scans that capture blood flow during a seizure, magnetoencephalography to map magnetic fields generated by brain activity, and in some cases, electrodes placed directly on or within the brain for more precise localization.
Surgery Offers the Best Chance of Seizure Freedom
For people whose seizures originate from a single identifiable brain region, surgery is the most effective treatment available. Roughly two-thirds of people with drug-resistant temporal lobe epilepsy (the most common type referred for surgery) become seizure-free after the affected tissue is removed. For other types of focal epilepsy outside the temporal lobe, the rate is around 50%.
Randomized trials make the advantage stark. In one landmark study, 58% of patients who underwent surgery were seizure-free at one year, compared to just 8% of those who continued medication alone. A later trial examining early surgical intervention found that 73% of surgical patients were seizure-free at two years, while none of the patients randomized to continued drug therapy achieved that outcome.
Success rates vary by the underlying cause. Brain tumors of certain types yield seizure freedom in 72% to 80% of cases after removal. Vascular malformations range from 70% to 91%. Focal cortical dysplasia, a developmental abnormality, sits around 58%. For children with severe epilepsy affecting one entire hemisphere, hemispherectomy (disconnecting or removing the affected half of the brain) achieves seizure freedom in 66% to 85% of cases.
Despite these numbers, surgery remains underutilized. Many people with refractory epilepsy wait years, sometimes decades, before being referred to an epilepsy center for evaluation.
Neuromodulation Devices
When surgery isn’t possible because seizures come from multiple areas or from regions that can’t safely be removed, implanted devices that deliver electrical stimulation to the brain or nervous system are an alternative. These don’t typically eliminate seizures entirely, but they can substantially reduce them.
Three devices are currently in use. Vagus nerve stimulation (VNS) involves a small generator implanted in the chest that sends regular electrical pulses to a nerve in the neck. Its effectiveness builds over time: about 28% of patients see at least a 50% seizure reduction at six months, rising to 52% by two years. Responsive neurostimulation (RNS) uses a device implanted in the skull that continuously monitors brain activity and delivers targeted stimulation the moment it detects the beginning of a seizure. It achieved a 66% seizure reduction at one year. Deep brain stimulation (DBS) targets a specific structure deep in the brain called the thalamus, with a 58% seizure reduction at one year.
A meta-analysis comparing the three devices found that RNS and DBS both outperformed VNS in the first year, though VNS narrows the gap over time as its effects continue to improve with longer use.
Dietary Therapies
The ketogenic diet, a high-fat, very-low-carbohydrate eating plan, has been used for epilepsy since the 1920s and remains a viable option for drug-resistant cases. In adults, pooled data show that roughly half of those who try the classic ketogenic diet achieve at least a 50% reduction in seizures, with about 13% becoming seizure-free.
The modified Atkins diet is a less restrictive version that’s easier to maintain long-term. In the largest observational study of 101 adults new to dietary therapy, 39% had at least a 50% seizure reduction on the modified Atkins diet, and 22% became seizure-free within three months. The classic ketogenic diet generally produces somewhat better results, but adherence is harder because of its strict ratios and limited food choices. Both diets require medical supervision and monitoring.
Newer Medications Still Play a Role
While the definition of refractory epilepsy hinges on two failed drug trials, that doesn’t mean all medications are equally ineffective. Cenobamate, one of the newest anti-seizure medications, has shown results that stand apart from other modern drugs. In clinical trials involving patients with uncontrolled focal seizures, cenobamate achieved seizure-free rates of 11% to 28% depending on the dose and study duration. Those numbers may sound modest, but they’re roughly double to triple the seizure-free rates seen with other newer medications, which generally fall below 5%.
Long-term data show that 16% of patients on cenobamate maintained seizure freedom for a median of nearly four years. Real-world studies have confirmed responder rates (at least 50% fewer seizures) of 63% to 70%, though complete seizure freedom in real-world settings is less common than in controlled trials.
The Toll of Uncontrolled Seizures
Living with refractory epilepsy affects far more than seizure frequency. Depression occurs in about 23% of people with epilepsy overall, roughly 2.7 times the rate in the general population, and the relationship with drug resistance runs in both directions: psychiatric conditions are associated with a four-fold increased risk of drug resistance. Anxiety disorders affect about 20% of people with epilepsy, with generalized anxiety being the most common form. Psychosis occurs in about 5% to 7%, and cognitive difficulties, particularly with memory and attention, are common complaints, often worsened by coexisting depression.
The most serious risk is SUDEP, or sudden unexpected death in epilepsy. The strongest risk factor is the frequency of generalized tonic-clonic seizures, the kind involving full-body stiffening and shaking. People who experience three or more of these seizures per year face a 15-fold increased risk of SUDEP, translating to up to 18 deaths per 1,000 patient-years. This is one of the most important reasons that achieving seizure control, through whatever combination of treatments it takes, carries real urgency.