LAUSR

Introduction The responsive neurostimulation (RNS) system is used for cases of medically refractory focal epilepsy in which curative resection is not possible. Uniquely, the device provides long term ambulatory neurophysiologic data including counts of intracranial seizure onset patterns. We sought to determine whether this data could be useful in predicting the response to starting an anti-seizure drug (ASD). Currently, there are no metrics available to accurately predict ASD response in a relatively short amount of time following its initiation. Methods We identified all patients with implanted RNS devices who were followed at the Yale Comprehensive Epilepsy Center between 2004 and 2016. 20 patients implanted with RNS had stable detection and stimulation settings suitable for further analysis. 21 instances of added medications during periods of stable detection settings were found. Medications that were not tolerated were not included. Daily detection outputs including “episode starts” and “long episodes” were compared before and after ASD initiation. Long episodes are electrocorticographic patterns that detection criteria for a provider-specified duration. Ratios of detection counts before and after each medication change were then compared with medication response outcomes. Medication success or failure was determined retrospectively based on provider documentation of improvement in seizures, whether or not the provider and patient chose to continue the medication long term, and patient-reported clinical efficacy including seizure diaries. Results Within 1 week of ASD initiation, relative to a 3-month baseline, long episode counts (p = 0.0008), but not total detection counts (p = 0.1733) were significantly different between instances of medication success and failure. Over a 3-month period after ASD initation, changes in long episode counts (p = 0.1258) and total detections (p = 0.6806) did not show a significant difference. Over the first week, a decrease in long episodes of 20% was an optimal cutoff point to determine efficacy. A decrease in long episodes of 20% or more predicted clinical efficacy with PPV = 100%. An increase in counts or a Conclusion In patients with RNS and stable detection settings, when new ASDs are started, a decrease in RNS long episode counts of 20% or more within the first week provides an early and accurate indication of individualized clinical efficacy.