Abstracts

Rationale: Our group has found that the production of interictal (InterPE) and postictal phonemic paraphasic errors (PostPE) and the time taken to correctly read aloud a test phrase immediately following seizure (postictal language delay; PILD) accurately lateralize language-dominant (DOM) temporal lobe epilepsy (TLE). Others including this group have also found improved diagnostic accuracy for individuals with normal intellectual ability (i.e., FSIQ≥90). The current study aims to further evaluate the effect of varying levels of overall intelligence on these three language indices.Methods: Sixty subjects (32 DOM TLE, 28 NDOM TLE) underwent inpatient video/EEG monitoring and had surgically confirmed temporal lobe epilepsy. We determined the presence of PostPE and, PILD times ≥60 seconds for 212 seizures, and presence of InterPE on the Boston Naming Test. Using FSIQ cut offs of ≥70 (N=60; 32 DOM, 28 NDOM), ≥80 (N=52; 27 DOM, 25 NDOM), and ≥90 (N=32; 14 DOM, 18 NDOM) each technique’s diagnostic usefulness was evaluated by computing sensitivity, specificity, positive predictive value (PPV), and negative predictive values (NPV) for each predictor.Results: Relative sensitivity, specificity, PPV, and NPV of each predictor for FSIQ≥70, ≥80, and ≥90 were, respectively, as follows; PILD (84%, 81%, 79%), (86%, 92%, 94%), (87%, 92%, 92%), and (83%, 82%, 85%); PostPE (94%, 93%, 93%), (64%, 68%, 72%), (75%, 76%, 72%), and (90%, 89%, 93%); InterPE (97%, 96%, 100%), (86%, 88%, 89%), (89%, 90%, 88%), and (96%, 96%, 100%). See Table 1.Conclusions: Diagnostic accuracy of InterPE, PostPE, and PILD was fairly high across all FSIQ levels (FSIQ≥70, ≥80, and ≥90), with only slightly improved classification accuracy as intelligence increased. This suggests the robustness of all three classification methods, and suggests that only modest improvements in diagnostic accuracy are obtained by restricting subjects to those with normal intelligence. Slightly improved accuracy with higher overall intelligence appears to result primarily from a reduction in number of false positives without a corresponding increase in false negatives. Results suggest that for individuals with FSIQ≥70 InterPE, PostPE, and PILD can be reliably employed in the pre-surgical evaluation of TLE to accurately differentiate between DOM and NDOM TLE.