B-28 Prognostic Utility of a Prediction Model for Memory Decline in Epilepsy Following Selective Amygdalohippocampectomy

Abstract

Abstract Objective It is important to evaluate risk for post-surgical cognitive decline in a systematic and reliable fashion due to the variable risk of memory decline following epilepsy surgery. Mathematical risk models have been developed to predict post-surgical memory decline (e.g., Stroup et al., 2003), but few studies have evaluated their prognostic utility. This study evaluated the predictive value of the Stroup et al., (2003) multivariate risk factor model in predicting postoperative verbal memory outcome among epilepsy patients who underwent left selective amygdalohippocampectomy. Method This retrospective study included 16 left-sided localization-related epilepsy surgery patients who underwent neuropsychological evaluation pre- and post-surgery. Patients were assigned to two risk groups: mild to moderate risk (15 to 60%) and severe risk (>61%) for postoperative memory decline, which was based on Stroup et al., (2003) criteria, suggesting that individuals with the highest risk had dominant hemisphere resections, no hippocampal sclerosis, normal Wada memory, and normal preoperative immediate and delayed memory scores. Memory decline was determined by pre- minus post-surgery memory test difference scores with 1.5 SD denoting a significant decline. Results The formula predicted postoperative memory decline with 75% accuracy for patients in the high risk group and no significant decline with 92% accuracy for patients in the mild to moderate risk group. Conclusion The results suggest that Stroup’s predictive formula for verbal memory outcome following epilepsy surgery may be useful for patients undergoing left selective amygdalohippocampectomy surgery, consistent with prior research suggesting prognostic utility with left temporal lobectomy patients (e.g., Johnson-Markve et al., 2009).