Abstracts

Reports on the diagnostic accuracy of structural MRI in detecting putative lesions in patients undergoing epilepsy surgery vary substantially, and are probably influenced by MRI protocols, types of lesions, and expertise. Using histopathology as the gold standard, we assess the overall sensitivity and specificity of structural MRI in an unselected cohort of patients undergoing epilepsy surgery, and provide post-tests probabilities for negative and positive studies., In a retrospective cohort of consecutive cases undergoing resective epilepsy surgery, we identified those who had histopathological tissue examination,ie., the gold standard. MRIs protocols contained a combination of general and epilepsy-specific sequences, and experts and non-experts performed MRI interpretation. The presence or absence of a putative lesion on structural MRI, using epilepsy specific protocols, was compared with the presence or absence of a histopathological lesion. We obtained sensitivity (true positive rate), specificity (true negative rate), positive and negative likelihood ratios, and post-test probabilities using a range of plausible pre-test probabilities. We also obtained 95% confidence intervals around each estimate. Factors influencing the diagnostic accuracy were assessed when the sample size allowed., Of 135 consecutive patients evaluated for epilepsy surgery, 117 had MRI and histopathological analyses. Histopathological abnormalities occurred in 104 patients (prevalence of 89%), of whom the MRI found a lesion in 93, yielding a false negative rate of 11%, and a sensitivity of 89% (95% CI = 83%, 95%). Of 13 patients with normal histopathology, the MRI showed a lesion in eight, with a false positive rate of 62%, and a specificity of only 38% (95% CI = 12%, 64%). The likelihood ratios for a positive and a negative MRI were 1.44 (low) and 0.3 (medium-high), respectively. Indicating that in this analysis, the MRI seems to be more useful when negative, because its false positive rate was too high. When the chance of finding a lesion is 50%, a positive MRI increases the probability to only 59%, and a negative MRI decreases is to 22%., In unselected patients undergoing epilepsy surgery Structural MRI has a sensitivity of almost 90% and a specificity of only 38%. The causes for the high false positive rate of MRI in our cohort are explored.
Reference:
Von OJ, Urbach H, Jungbluth S et al. Standard magnetic resonance imaging is inadequate for patients with refractory focal epilepsy. J Neurol Neurosurg Psychiatry 2002;73:643-647.,