Abstracts

Rationale: Epileptic spasms commonly recur in children with Infantile Spasms Syndrome (ISS)–more than a third of those who initially remit will relapse. Single center studies suggest some clinical features such as epileptiform discharges are associated with relapse, though the accuracy of these features to predict relapse is understudied. Here, we calculate the sensitivity and positive predictive value (PPV) of clinical and electroencephalogram (EEG) variables for relapse. We also propose a clinical algorithm to identify those most at risk.

Methods: We performed chart reviews of children with ISS in the Rare Epilepsies of New York City (RENYC) database who had a date of birth from 2010-2013 as well as chart review of patients with ISS at Weill Cornell Medicine from 2009-2018. We included children who had complete response to first line treatment within 14 days and who had a durable response of at least 28 days. We coded demographic variables, EEG findings at diagnosis and at follow-up, imaging findings, and whether patients relapsed. We excluded children followed for less than four months. We were motivated to include epileptiform discharges as a predictor because of prior work demonstrating association with relapse and high inter-rater reliability among EEG readers.

Results: Fifty-one children met study criteria. These children were diagnosed at a median of 6.2 months [Interquartile interval (IQI) 5.1-8.5 months, range 1.7 – 18.7 months] and were followed for a median of 25 months [IQI 14.5 – 42.9 months, range 5.5 – 137.8 months], with follow-up EEGs performed at a median of 33 days [IQI 26 - 42 days, range 0 – 384 days]. Thirty-three (65%) children were kept on some sort of anti-seizure medication or diet after initial treatment, with levetiracetam (22%), vigabatrin (20%), and topiramate (18%) being the most common. Eighteen children (35%) relapsed, at a median of 130 days [IQI 76 – 147 days, range 41 – 248 days]. Presence of epileptiform discharges had a large effect size in favor of relapse, though with broad confidence intervals (CI) (Hazard Ratio (HR) 4.0; 95% CI [0.76 - 20.9]; Table 1). To increase PPV for identifying those who will relapse, we examined those with epileptiform discharges at follow-up and found three variables with high sensitivity and/or PPV for relapse: slowing on follow-up EEG (79% sensitivity), developmental delay before treatment (80% sensitivity), and deep grey matter injury on imaging (60% PPV). We used this information to create a clinical algorithm to identify those who will relapse by assigning a point for the presence of each variable to each individual. Those with three points had the highest risk of relapse (75%) whereas those with zero points had a 0% risk of relapse (Table 2).

Conclusions: Presence of epileptiform discharges at follow-up was highly sensitive for identifying those who will relapse but had poor PPV. Additional consideration of slowing on follow-up EEG, developmental delay before treatment, and deep grey matter injury can allow physicians to identify who will relapse with higher PPV. The study was limited by small sample size.

Funding: Please list any funding that was received in support of this abstract.: N/A.