Epilepsy surgery: non-invasive versus invasive focus localization. What is needed from a neuropsychological point of view

Interrater reliability among epilepsy centers: multicenter study of epilepsy surgery

PURPOSE

To measure the interrater reliability of presurgical testing and surgical decisions among epilepsy centers.

METHODS

Seven centers participating in an ongoing, prospective multicenter study of resective epilepsy surgery agreed to conform to a detailed protocol regarding presurgical evaluation and surgery. To assess quality assurance, each center independently reviewed 21 randomly selected surgical cases for preoperative study lateralization and localization, and surgical decisions. Interrater reliability was assessed by using intraclass correlation coefficients (ICCs), validated for use with multiple raters, and calculated in a two-way random model based on absolute agreement.

RESULTS

Agreement for ICC values: > or = 0.75, excellent; 0.60-0.74, good; 0.40-0.59, fair; < or = 0.39, poor. One center was excluded for missing data. Agreement was excellent for extracranial EEG lateralization (0.8039), magnetic resonance imaging (MRI) lateralization (0.9521) and localization (0.9130), Wada lateralization (0.9453), and intracranial EEG localization (0.7905). Agreement was good for extracranial EEG localization (0.7384) and neuropsychological testing lateralization (0.7178) and localization (0.6891). Consensus about the decision to perform intracranial monitoring was fair (0.5397), in part reflecting one center's tendency toward intracranial monitoring. Overall agreements on whether to perform surgery (0.8311) and specific surgery recommended (0.8164) were excellent.

CONCLUSIONS

High interrater reliability among six epilepsy centers was present for interpretation of most components of presurgical testing. Although consensus for the decision to perform intracranial monitoring was only fair, agreements for the ultimate decision about resective surgery and specific choice of resection were excellent. We believe that this study demonstrates the feasibility of implementing multicenter protocols for neurologic management, especially those involving localization, as well as protocols combining study results with clinical decision making.

Is refractory epilepsy preventable?

About a third of the patients diagnosed with epilepsy will not be fully controlled with antiepileptic drugs (AEDs), and many of them will have frequent and disabling seizures. These patients will undergo multiple drug trials, most often without complete seizure remission. Moreover, refractory epilepsy is associated with increased morbidity (from seizures and medications), social isolation, unemployment, and overall reduced quality of life. There is evidence that refractory epilepsy can be a progressive disorder, which, if controlled early, might never develop into a full syndrome with all of its associated sequelae. The difficulty lies in identifying at an early stage patients who are likely to progress to intractability. No currently known markers enable clinicians to make this identification with confidence. Advances in pharmacogenomics and our understanding of pharmacologic responsiveness in epilepsy may change this situation. Even now, we are able to identify many patients with a poor prognosis earlier than before, particularly in the pediatric population, in which syndromic classification may provide an approach to predict intractability. The early initiation of aggressive therapy may improve outcome and overall quality of life.

Prediction of verbal memory loss in individuals after anterior temporal lobectomy

PURPOSE

Decreased memory function represents the area of greatest neuropsychological morbidity after anterior temporal lobectomy (ATL), particularly for left ATL candidates. We wished to identify easily derived demographic and neuropsychological predictors of risk of pre- to postoperative memory decline using only information available preoperatively.

METHODS

We assessed decline in memory as measured by the California Verbal Learning Test (CVLT) by deriving multiple regression equations using the following measures as independent variables: age at onset, chronological age at time of surgery, sex, Full Scale IQ (FSIQ), level of education, and preoperative memory scores. In all, 203 patients (93 males, 110 females), undergoing ATL (107 left, 96 right) with preoperative and 6-month postoperative testing, were examined.

RESULTS

The combination of age, FSIQ, sex, side of surgery and preoperative score was highly predictive (p-values < 0.0001) of postoperative memory scores. Higher postoperative scores were associated with higher preoperative score, younger chronological age, higher FSIQ, female sex, and right side of resection. Reliable change index (RCI) values were used to estimate meaningful decline on the total score across five trials. Logistic regression analysis showed preoperative score and age to be predictors of RCI decline for left-sided resections. Sensitivity of decline (> or =90th centile RCI) prediction was 56%, and specificity was 95%. Validation in 30 patients from a separate population of patients undergoing left ATL produced similar figures.

CONCLUSIONS

The derived regression equations can accurately predict verbal memory decline on a list-learning task in approximately 50% of individual patients undergoing ATL, and false-positive prediction errors are very rare.