Temporal Lobe Epilepsy and Volumetric MRI in Clinical Trials

Temporal Lobe Epilepsy

Temporal lobe epilepsy (TLE) localized to the hippocampus (HC) known as hippocampal sclerosis (HS) is one of the more common types of localizing epilepsy. It is typically refractory to medical therapy.

Structural magnetic resonance imaging (MRI) imaging of HS relies on the detection of hippocampal atrophy, gliosis manifested by abnormal signal intensity on T2-weighted images and loss of internal architecture. Identifying patients that may benefit from surgical resection, laser thermal ablation or surgically implanted stimulator devices first requires accurate diagnosis.

A positive diagnosis is established by concordance between scalp electroencephalogram (EEG) monitoring and structural MRI findings. A patient is then evaluated for one of the interventions mentioned above after obtaining neuropsychiatric testing to establish baseline cognition and successful screening for language and memory lateralization to exclude resection or ablation of eloquent brain ipsilateral to hippocampal pathology. With concordance between scalp EEG and structural MRI findings, a seizure free outcome to
temporal lobe resection ranges between 60 and 80%.

In the absence of concordance between scalp EEG and structural MRI, further testing with invasive EEG monitoring, magnetoencephalography (MEG) or positron emission tomography (PET) scanning may be required. With discordance between scalp EEG and structural MRI findings, a seizure free outcome to temporal lobe resection is less than 60%.

Diagnostic Challenges in Patient Selection for Clinical Trials

The diagnosis of TLE is problematic in the absence of definitive EEG or structural MRI abnormalities. Hippocampal atrophy is the imaging biomarker for HS. It is generally accepted that an experienced neuroradiologist can detect hippocampal volume loss on the order of 15%. This requires that a radiologist set a subjective threshold for hippocampal size based on their experience with normal variation upon review of hundreds if not thousands of images. When a patient has bilateral HS, a relatively less common subset of

TLE, even an experienced neuroradiologist can have difficulty suggesting a diagnosis. Mild bilateral hippocampal atrophy provides a unique challenge to perception for all interpreters of cross sectional images, who instinctively use asymmetry to perceive abnormalities based on size. While the adjunct findings of gliosis or loss of internal architecture of the HC can be helpful, they are not true biomarkers and are less specific than atrophy in diagnosis.

The quantitative analysis of hippocampal volumes using FDA-cleared software provides a more accurate assessment of both unilateral and bilateral HS than subjective visual assessment of structural MRI images alone.

Volumetric MRI Establishes a Quantitative Biomarker for Hippocampal Sclerosis

Hippocampal volumes can be quantified using commercially available software products which compare left and right hippocampal volumes to an age-matched database. This quantitative biomarker can then be used to set a numerical threshold for inclusion in clinical trials and/or to place other- wise qualified subjects into stratified categories based on MRI severity of hippocampal volume loss.

Quantitative volumetric software also creates a measure of internal asymmetry between left and right hippocampal volumes known as the asymmetry index (AI). The AI is also a biomarker for HS which correctly classified 94% of patients with left versus right TLE in a single study. A mean hippocampal asymmetry of 25% was present in TLE subjects in this same study. An AI of 25% falls below the 5th percentile or corresponds to a 95% confidence level.

While a subjective estimate of a 10-15% asymmetry between hippocampi may be an unrealistic detection goal for even an experienced neuroradiologist, this approaches a 1.5 standard deviation difference or a 90% confidence level. Such is the case when individual hippocampal volumes deviate from the mean in different directions by less than one standard devia- tion. Conversely, in patients with bilateral HS, the AI is not particularly useful.

The quantitative comparison of each hippocampus with age-matched controls can establish the bilateral diagnosis along with a stratification of severity for each side.

Patient Categorization in TLE Clinical Trials

In patients with discordant scalp EEG and structural MRI findings, the volumetric MRI with quantitative analysis can more accurately identify patients for study selection and stratify disease severity with an objective measurement of hippocampal volume loss and asymmetry.

In patients with negative volumetric and structural MRI studies, two scenarios may exist. In the first, the patient may have HS based on scalp EEG or invasive EEG monitoring and abnormal MEG or PET studies. This would be analogous to the SWEDD classification (scans without evidence of dopamine deficiency) in patients with Parkinsonism. In retrospect, many of the patients with Parkinsonian symptoms and the SWEDD design ation ma y not hav e ha d a neurodegenerative process but instead were incorrectly classified. Some patients, on long term follow-up, were later categorized with non-degenerative Parkinsonism (drug-induced or vascular) or essential tremor.

If patients with negative volumetric and structural MRI studies are to be included in TLE clinical trials, their data should be characterized as such so as not to skew clinical trial results negatively. This would be the second scenario where the diagnosis established without volumetric or structural MRI abnormalities is circumspect.

Volumetric MRI as a Predictor of Therapeutic Response

Volumetric MRI with quantitative analysis provides a much more sensitive imaging assessment than structural MRI alone as it can describe both the severity of HS and predict memory outcome following surgical resection.

Temporal lobe resection in TLE has become a more precisely localized surgical procedure as the quality of structural and functional imaging has improved. In a recent study, the magnitude of the AI correlated linearly with the memory outcome in patients with temporal lobe resection. The greater the magnitude of asymmetry, especially in patients with left smaller than right hippocampal volume loss, the less likely they were to experience memory decline. A few patients experienced memory improvement. Conversely, patients with smaller degrees of asymmetry experienced some memory decline following surgical resection.

Stereotactic laser thermal ablation of the amygdala and hippocampal head using a surgically implanted catheter is a minimally invasive technique associated with a lower risk of cognitive decline. It would be interesting to study the param- eters obtained in volumetric MRI to determine if therapeutic response stabilized the progression of volume loss compared with TLE patients randomized to no treatment or medical treatment. This could have long term prognostic implications as a surrogate endpoint in seizure free survival.

Responsive stimulation using an implanted device offers a new approach that senses electrical activity in a seizure focus and delivers a localized electrical pulse to disrupt the seizure activity. These devices can stimulate up to two spatially discrete seizure foci, either bilateral hippocampi or a unilateral hippocampus with a separate cortical focus. As these patients would be excluded from volumetric MRI post-implantation, it would be interesting to see if their volumetric datapoints correlated with the therapeutic endpoints after device implantation.

---

Volume 5, Issue 4: Guidance For Sponsors: Temporal Lobe Epilepsy and Volumetric MRI in Clinical Trials

Originally written by legacy Intrinsic Imaging Medical Director