Abstracts

Rationale: Although routinely used in clinical practice where available, the appropriate role of functional imaging tests such as FDG-PET, ictal SPECT, and MEG based source imaging (MSI) in epilepsy surgery decision-making is unclear. These tests are expensive and a burden to obtain for both patient and caregiver. Whether these tests should be used to influence any aspect of patient selection, decisions of whether or where to place intracranial electrodes and surgical resection needs to be determined such that actual clinical utility dictates treatment. The stakes are high for influences on an intervention that can be effectively a cure of what otherwise nearly always remains a lifelong disabling disease. Methods: Analysis was based on test sensitivity / specificity and surgical outcome data obtained in a prospective nonlesional epilepsy surgery project performed at UAB between 2001 and 2006. Values for health related outcome states were obtained from published quality of life (QOLIE) measures. The mean values of the following health states were used: utility of cure (uCure, completely free of disabling seizures)=0.97, uCure + disability (from surgery)=0.80, utility of continued disabling seizures (uSeizures)=0.70, and uSeizures + disability=0.60. The base probability of harm from surgery (pHarm=0.05) was also based on literature reports. The probability of cure (pCure) for both medical and surgical treatment, and the probability of added cure because of test influence (test effect) were varied for one- and two-way sensitivity analyses using TreeAge Pro 2009 (www.treeage.com). Treatment strategies included 1) continued medical management, 2) treat all (surgery), 3) treat all with added test effect (test used to influence localization of surgery and outcome), and 4) sort based on test, then treat-either surgery (includes test effect) or medical. Results: A “treat all” context dominates because the risk/benefit ratio is so strongly skewed toward the large QOLIE gain achievable with elimination of disabling seizures. As a result, without any other contribution a test sort role to select patients (surgical versus continued medical treatment) is not a preferred strategy unless sensitivity is greater than 98%. Test specificity only begins to effect the decision strategy if pHarm ≥ 50%. For any sort role to be included in the strategy based on estimated risks and benefits, a test effect increase in pCure is necessary. For example, a test effect improvement in seizure-free outcome ≥19% is necessary for MSI (sensitivity=72%) to play a role in patient selection. However, as long as the test contributes to any increase in pCure, all candidates should have the test, and if possible, all should be treated with surgery. Conclusions: Without taking cost into consideration, functional imaging tests (alone) should not be used to deny surgical treatment to any patients. However, if adding the test affects any increase in proportion of seizure-free outcomes, using the test in all possible candidates, followed by surgery, is preferred.