Stopping AEDs After Four Seizure-Free Years, With One Ambiguous EEG
A single patient, four years seizure-free, asking to stop his antiepileptic drug. The guideline math on recurrence risk is clear — his own repeat EEG is what complicates it.
P.N., a 27-year-old man, was diagnosed with focal epilepsy at nineteen after two witnessed seizures during his first year of college, and has been fully seizure-free for four years on lamotrigine 200mg twice daily. He recently started a new job as a paralegal, a career change he worked hard for after finishing his degree part-time, and has told his neurologist directly that he'd like to try stopping the medication — not because of any side effect he can point to, but because he'd like to stop thinking of himself as someone who takes a seizure drug, now that four years have passed without incident.
His neurologist ordered a repeat EEG specifically to inform the withdrawal decision, following the practice-guideline logic that a normal study meaningfully lowers recurrence risk. The result came back not normal, but not clearly abnormal either: rare left temporal sharp waves, present but infrequent enough that the reading neurologist flagged it as a borderline finding rather than a clear-cut epileptiform abnormality. His neurologic exam is normal, his MRI from diagnosis showed no structural lesion, and he has never had a seizure while driving or in a way that suggests a higher-risk pattern — but an epileptiform EEG is the one predictor the AAN/AES practice advisory update (Gloss et al., 2021) found any real signal for, having judged the evidence insufficient to support or refute every other candidate factor. The size of that signal is easy to overstate: the advisory grades it on Class III pediatric studies at low confidence and states plainly that in adults there is no evidence above Class IV. P.N. is twenty-seven, and this EEG isn't unambiguously clean.
Once he heard the result, he asked directly whether a few rare sharp waves on one recording should really be enough to keep him on a daily medication he's been asking to stop for over a year. Nobody in the room could give him a number.
After the repeat EEG comes back borderline
I'd hold off. Of all the candidate predictors of recurrence after withdrawal, an epileptiform EEG is the only one the AAN/AES practice advisory update (Gloss et al., 2021) singled out as carrying any real signal at all — everything else it looked at, it judged there was insufficient evidence to support or refute. His is a borderline reading, not a clean one, and 'borderline' on the one variable that survived that review isn't the same as 'reassuring.'
A year of him asking to stop doesn't change what the sharp waves mean — the advisory's recurrence-risk estimates were derived from EEG findings, not from how long or how consistently a patient has wanted off the drug.
You're right that epileptiform activity is the one predictor that survived that review — I'm not disputing that. But I'd push back on how much weight it carries for him specifically. Gloss and colleagues graded the epileptiform-EEG signal on Class III pediatric studies at low confidence, and said plainly that in adults there is no evidence above Class IV. He's twenty-seven. We're extrapolating a low-confidence pediatric finding onto an adult and calling it the reason to keep him on a drug. The same advisory puts the absolute numbers at roughly 15% recurrence over 24 to 60 months in adults who taper versus 7% in those who don't — real, but a long way from expected recurrence, and he has four full years of actual seizure freedom behind him, which is evidence too, not just a number on one recording.
He's also not asking to stop today and never think about it again — he's asking for a real conversation about a real, guideline-supported option, at a point in his life where the medication's ongoing cost to how he sees himself is genuinely part of the calculation, not something to set aside because one EEG came back ambiguous.
There's a way to take both of those seriously without picking one over the other outright. A slow taper — months, not weeks — with driving restricted during the taper itself and for a defined interval afterward, and a clear seizure action plan given directly to him and to whoever he lives with, addresses the actual mechanism of risk here: most recurrences after withdrawal happen either during the taper or in the months immediately following it, not years later.
If he tapers and stays seizure-free through that window, the borderline EEG becomes a reasonable data point in a genuinely favorable overall picture. If he doesn't, we've learned something real about his risk with the smallest possible cost — rather than either overriding a real guideline predictor to satisfy a preference, or holding him on a drug indefinitely because of one ambiguous study he's entitled to ask us to weigh honestly.
Agreed: a slow taper over several months, driving restricted through the taper and for a defined period afterward, and a written seizure action plan reviewed with him and his roommate before the first dose reduction.
Not agreed: the Epileptologist's underlying read — that the borderline EEG alone argues against withdrawal at all — was not adopted as the final decision, but was not treated as wrong either; it was named directly as the position the taper protocol is designed to hedge against, not one the team resolved by outvoting it.