Authors :
Presenting Author: Katherine Walsh, BS – Massachusetts General Hospital
Wen Shi, PhD – Department of Neurology – Massachusetts General Hospital; Pauline Brandon Bravo Bruinsma, BSc – Division of Epilepsy and Clinical Neurophysiology – Boston Children's Hospital; Dana Shaw, BS – Graduate Program in Neuroscience – Boston University; Darcy Krueger, MD, PhD – Division of Neurology, Department of Pediatrics – Cincinnati Children's Hospital Medical Center; Mustafa Sahin, MD, PhD – Division of Epilepsy and Clinical Neurophysiology – Boston Children’s Hospital; Hope Northrup, MD – Division of Medical Genetics, Department of Pediatrics – University of Texas Medical School at Houston; Joyce Wu, MD – Division of Neurology & Epilepsy – Ann & Robert H. Lurie Children’s Hospital of Chicago, Northwestern University Feinberg School of Medicine; Martina Bebin, MD – Department of Neurology – University of Alabama at Birmingham; Mark Kramer, PhD – Department of Mathematics and Statistics & Center for Systems Neuroscience, Boston UniversityDepartment of Mathematics and Statistics & Center for Systems Neuroscience – Boston University; Jurriaan Peters, MD, PhD – Division of Epilepsy and Clinical Neurophysiology – Boston Children's Hospital; Catherine Chu, MD, MMSc, MA – Department of Neurology – Massachusetts General Hospital
Rationale: Approximately a third of infants with tuberous sclerosis complex (TSC) will develop infantile spasms (IS) and pre-treatment with vigabatrin improves outcomes but also introduces risk of permanent retinal toxicity. Improved detection of risk for future IS is required to optimize outcomes and minimize risk for these children. Interictal epileptiform activity may precede the presentation of epilepsy in infants with TSC. In children and adults with epilepsy, spike ripples, the combination of interictal spikes with high frequency oscillations (80-500 Hz), has improved specificity for epileptogenic cortex compared to spikes or ripples alone. We evaluated the performance of focal spike ripple rate to predict future IS in pre-symptomatic infants with TSC.
Methods: We analyzed the pre-symptomatic EEGs collected from a prospective multicenter observational study of children with TSC and follow up for at least 36 months. Infants were grouped into those who developed future IS and those who did not. Non-rapid eye movement Stages 2 and 3 sleep were identified from the most recent pre-symptomatic EEG (mean 20.8 mins, range 6-59 mins) and analyzed using a semi-automated spike ripple detector. Candidate spike ripple detections were classified as true spike ripples or false detections by blinded expert review. The spike ripple rate in the most active EEG channel for each infant was compared between groups using a generalized linear model with age as a covariate. Among children who developed IS, we evaluated whether duration from the time of the analyzed EEG until the onset of IS impacted spike ripple rate using a generalized linear model accounting for age. The optimal performance threshold to classify infants into those with and without future IS was computed and the accuracy of the classifier was evaluated.
Results: A total of 29 infants with TSC with pre-symptomatic EEGs were included, including eleven who developed future IS (mean time from EEG to IS onset 48 days, range 6-106) and eighteen who did not. Among the eighteen children who did not develop IS, six developed future focal epilepsy (mean time from EEG to seizure onset 96 days, range 24-259 days) and twelve did not develop any seizures over the follow-up period. After controlling for age (beta=0.31; p=0.002), the expert-validated spike ripple rate was higher in children who developed IS compared to those who did not (beta =1.96; p=0.020). Duration until IS onset did not impact spike ripple rate (p=0.9). Using a threshold of 0.07 expert-validated spike ripples per minute, future IS was classified with 72% accuracy, with a specificity of 89% and sensitivity of 45%.
Conclusions: Spike ripples in pre-symptomatic EEGs may identify TSC infants who will develop future IS.
Funding:
NIH NINDS R01NS119483, NIH U01-NS082320, P20-NS080199