Abstracts

Rationale: Tuberous Sclerosis Complex (TSC) is a rare genetic neurocutaneous disorder strongly associated with epilepsy. It is refractory to medication in over 65%, and poorly controlled epilepsy is associated with adverse neurodevelopmental outcome. In those children, epilepsy surgery should be considered, but due to the multilesional nature of the disorder, identification of the epileptogenic zone (EZ) is a formidable challenge. While magnetic resonance imaging (MRI) provides anatomical insight in the many candidate lesions, limited data exists on which structural MRI features mark the EZ in young children. Methods: Preoperative MRI of 26 children with TSC who underwent epilepsy surgery before age 3 years at 5 collaborating sites of the TSC Autism Center of Excellence Research Network (TACERN) were analyzed by two experienced neuroradiologists, blinded to surgical approach and outcome data. In ten selected lesions in each subject, the neuroradiologists independently scored for presence or absence of three tuber characteristics (large affected area, calcification, cyst-like properties) and three features traditionally associated with focal cortical dysplasia (FCD) (cortical malformation, gray-white matter junction blurring, transmantle sign). Next, based on their expertise, the radiologists identified and rank-ordered the three lesions most suspect for the EZ. Finally, they were unblinded to electroencephalography (EEG) data on ictal localization of the seizure targeted in the planned surgery, and could re-order the three lesions. We examined which individual or combination of features was associated with the EZ by comparing their localization with the resection site in post-operatively seizure-free patients. Accuracy in localizing the EZ and intra-observer variability (IOV) were also calculated.  Results: Resected lesions in seizure-free children, reflecting the EZ, more often affected a large area (p=0.044). Additionally, epileptogenic lesions scored higher on harboring calcifications (p=0.027) and one or more tuber features (p=0.002). Large affected area and tuber features as a cluster showed moderate sensitivity (50.0% and 58.4%, respectively) and specificity (65.6% and 64.0%, respectively). Calcification, however, demonstrated low sensitivity (19.6%) but high specificity (92.0%). There was no association between epileptogenic lesions and the other features studied, including cyst-like properties, cortical malformation, gray-white matter junction blurring, and transmantle sign. Moreover, no combination of any of the MRI markers was more prevalent in epileptogenic lesions as compared to non-epileptogenic lesions. IOV showed a moderate agreement (kappa = 0.59) between the reviewers in assessment of all MRI features and an almost perfect agreement in assessment of the features large affected area (kappa=0.94), calcification (kappa=0.89), and tuber characteristics as a cluster (kappa=0.860). Compared to review of structural MRI alone, the addition of EEG data led to an increase in identification of the resection site in seizure-free (from 47.8% to 88.6%) and non-seizure-free children (from 50% to 59%). Thus, with EEG, discrimination between epileptogenic and non-epileptogenic lesions significantly improved (p=0.009). Conclusions: Epileptogenic lesions in TSC more often affect a large area and more often contain calcifications than non-epileptogenic lesions. FCD features were not distinctive for the EZ. When combined with EEG data on the ictal localization, the assessment of MRI features resulted in increased accuracy in localization of the epileptogenic tuber.  Funding: No funding