Sample case: left mesial temporal hypothesis

Presentation

A 34-year-old right-handed woman with drug-resistant focal epilepsy. She had a single prolonged febrile seizure lasting about 25 minutes at 18 months of age, then developed normally. Afebrile seizures began at age 14.

She has tried levetiracetam (stopped for irritability), lamotrigine and lacosamide, and currently takes lamotrigine with lacosamide at maximum tolerated doses. She has three to five focal impaired-awareness seizures a month and one focal to bilateral tonic-clonic seizure in the past year. She works as a schoolteacher and reports word-finding difficulty and trouble remembering conversations. There is no family history of epilepsy, and no history of encephalitis or head injury.

Semiology

Six typical seizures were recorded during a 5-day video-EEG admission. All had the same sequence:

1. Aura (about 10 seconds). A rising epigastric sensation, sometimes with déjà vu. She can press the event button and describe it.
2. Loss of awareness with behavioral arrest and staring.
3. Oroalimentary automatisms (lip-smacking and chewing) from about 20 seconds.
4. Left-hand fumbling automatisms with dystonic posturing of the right arm.
5. Post-ictal phase. Left-handed nose-wiping, then marked word-finding and naming difficulty for 5 to 10 minutes. Seizures lasted 60 to 90 seconds.

In two of the six seizures the aura also included a tight, choking feeling in the throat and hypersalivation before awareness was lost.

Pause and think Which features lateralize, which localize, and is anything atypical for a pure mesial temporal onset?

The left-hand automatisms with right-arm dystonia, post-ictal aphasia and left-handed nose-wiping all point to the left hemisphere, which is dominant for language. The epigastric aura, déjà vu and early oroalimentary automatisms localize to the mesial temporal lobe.

The throat constriction and hypersalivation are the atypical part. They are classic insulo-opercular symptoms, and oroalimentary automatisms can themselves be generated by insulo-opercular cortex. Together they raise the possibility of temporal plus epilepsy, with the insula or orbitofrontal cortex involved early.

Scalp EEG

Interictal: frequent left anterior temporal sharp waves (maximal at FT9 and F7), with rare independent right temporal sharp waves (under 10%). There was intermittent left temporal rhythmic delta.

Ictal: in five of six seizures, rhythmic 5–6 Hz theta built up over the left temporal chain within 15 seconds of clinical onset. In the sixth, the onset was poorly localized, with diffuse attenuation before left temporal rhythmic activity emerged. In every seizure the aura began before any scalp EEG change.

Imaging

3T MRI (epilepsy protocol): no hippocampal atrophy or signal change, and symmetric hippocampal volumes. On re-review, a question of subtle gray-white blurring in the left temporal pole was raised but not confirmed.

FDG-PET: left anterior and mesial temporal hypometabolism, most marked in the temporal pole, with mild extension into the left anterior insula.

Language fMRI: left-hemisphere dominant.

Neuropsychology

Verbal memory was impaired, with poor delayed recall of word lists and stories. Confrontation naming was low average. Visual memory, attention and executive function were within normal limits. This pattern fits dominant (left) temporal dysfunction, and it means verbal memory and naming are at risk with left temporal surgery.

Hypothesis

The conference agreed on a left mesial temporal (amygdala-hippocampal) onset as the leading hypothesis. The alternatives were a temporopolar onset and temporal plus epilepsy with early anterior insular or orbitofrontal involvement, prompted by the laryngeal symptoms, the hypersalivation and the PET extension into the anterior insula.

She went to SEEG rather than straight to surgery for three reasons. The MRI is negative. The surgery is in the language-dominant hemisphere with verbal memory already affected. And the answer changes the operation: a mesial onset could be treated with laser ablation, a temporopolar onset with an anterior temporal lobectomy, and true temporal plus epilepsy with a tailored, extended resection.

Pause and think What questions does the implantation need to answer?

1. Is seizure onset mesial (entorhinal, hippocampal or amygdalar), temporopolar, or lateral neocortical?
2. Are the insula and orbitofrontal cortex part of the onset, or recruited later as propagation?
3. How far back does mesial involvement extend?
4. Where is language cortex in the basal temporal lobe, so treatment can stay clear of it?
5. If surgery isn’t possible or doesn’t work, does the seizure network reach a thalamic neuromodulation target?

Each lead in the plan is there to answer one of these.

Implantation plan

Ten orthogonal depth electrodes were placed in the left hemisphere: the team’s standard mesial/lateral temporal scheme, plus an orbitofrontal lead (OR’) for the temporal plus question. The deep contacts of each lead reach the target named below, and the contacts along the way sample everything the trajectory crosses. That is how the insula is covered by T’ and the temporal pole by J’, without dedicated leads.

SEEG findings

Antiseizure medications were tapered, and five habitual seizures were recorded over 8 days.

• Onset: every seizure began with pre-ictal spiking, then low-voltage fast activity (about 25–30 Hz) in the deep contacts of B’ (hippocampal head) and E’ (anterior entorhinal cortex), with the hippocampal body contacts of C’ a second or two later.
• Early spread: the amygdala (A’) within 3–4 seconds, then the temporal pole (I’, and the entry contacts of J’) at about 8 seconds. The pulvinar contacts of U’ were recruited at about 10 seconds. The patient pressed the event button for her epigastric aura at about 5 seconds.
• Late spread: the inferior insula contacts of T’ at about 15 seconds, coinciding with the throat constriction in the two seizures that had it, and its centromedian thalamus contacts shortly after. The fusiform contacts of F’ were recruited later still. Orbitofrontal cortex (the deep contacts of J’, and OR’) was involved in only one seizure, at about 30 seconds.
• Interictal: frequent spikes in B’, E’ and A’, and rare spikes in I’.

Stimulation: 1 Hz stimulation of the mesial contacts of B’ reproduced her habitual epigastric aura with déjà vu. 50 Hz stimulation of the posterior basal contacts of F’ caused naming arrest, identifying the basal temporal language area.

Pause and think Does the insular or thalamic involvement change the surgical plan?

No. The insula was recruited about 15 seconds after onset in the hippocampus and entorhinal cortex, after the aura had already begun. That makes it a propagation site rather than part of the epileptogenic zone, and the orbitofrontal cortex was almost never involved. The thalamus was recruited as expected for a mesial temporal seizure. The epileptogenic zone is confined to the left mesial temporal structures, and the stimulation results mark language cortex posterior and basal to them.

The thalamic recordings still matter: early pulvinar recruitment means pulvinar stimulation is a reasonable fallback if ablation doesn’t control her seizures.

Surgery and outcome

The team weighed a standard left anterior temporal lobectomy against a more selective approach. Because the epileptogenic zone was mesial, the hemisphere is language-dominant, and verbal memory and naming were already vulnerable, she had stereotactic laser ablation of the left amygdala, hippocampal head and anterior hippocampal body, and entorhinal cortex. The ablation stayed anterior to the basal temporal language site found on stimulation.

She had two brief isolated auras in the first three months and has been seizure-free since, at 18 months of follow-up (Engel I), so the pulvinar fallback has not been needed. Lacosamide was tapered off. At one year, naming was unchanged and verbal recall had declined slightly, a change she did not notice day to day.