Who will use epilepsy surgery nomograms, and why?

240 www.thelancet.com/neurology Vol 14 March 2015 Advances in neurodiagnostic technology and mi crosurgery have greatly improved the safety and effi cacy of surgical treatment for epilepsy in recent years. Results from two randomised controlled trials have shown the superiority of surgical treatment over continued drug treatment for patients with medically refractory temporal lobe epilepsy. A practice parameter produced by the American Academy of Neurology recommended surgery as the treatment of choice for this type of epilepsy, and acknowledged that similar results could be obtained for drug-resistant neocortical epilepsy. More recent studies have shown that surgical benefi t is long-lasting for most patients, and that surgical mortality and complications are rare. Nevertheless, surgical treatment for epilepsy remains one of the most underused accepted therapeutic interventions in medicine. It has been estimated that less than 1% of all patients in the USA who meet the International League Against Epilepsy criteria for drug-resistant epilepsy— persistence of disabling seizures despite two appropriate antiseizure drug treatments—are referred to epilepsy centres, where candidacy for surgery can be ascertained. In fact, referral to epilepsy centres in the past several decades has decreased, and the average duration of epilepsy for those who are referred for surgery has increased to more than 20 years, often too late to avoid irreversible disability. The fi rst randomised controlled trial and the American Academy of Neurology practice parameter seem to not have stimulated earlier referrals. In an eff ort to improve identifi cation of potential candidates for surgery, Lara Jehi and colleagues constructed two nomograms to predict surgical outcome based on information collected from 846 patients before referral to the Cleveland Clinic (Cleveland, OH, USA). They validated these results with data collected from 604 patients before referral to four other epilepsy centres in Europe, Latin America, and the USA. They conclude that these nomograms could be an innovative and practical instrument for individualised prediction of seizure outcome after surgical treatment for epilepsy. This work shows an increasing interest in applying statistical predictive modelling to prognosis of diseases and treatments. Although far from perfectly predictive, the nomograms seem to be a step towards the identifi cation of patients who are most likely to benefi t from surgery. Because they are based on data available before the presurgical assessment, the nomograms do not include interictal or ictal electroencephalogram, PET, ictal SPECT, magnetoencephalography, intracranial recording, or many other advanced techniques commonly used in epilepsy centres to locate an epileptogenic region. Furthermore, they do not account for the potential for surgical intervention to improve health-related qualityof-life, or the risk of such an intervention introducing an unacceptable new neurological defi cit. These factors are all essential in establishing whether physicians at the epilepsy centre will recommend surgery, and whether the patient will accept this recommendation. An important question, therefore, is who will use such nomograms and for what purpose? If such nomograms are used in epilepsy centres, they could be applied at the time of the initial visit, to help a patient who might be ambivalent about surgical treatment to decide whether or not to undergo a presurgical assessment. Alternatively, such nomograms could be widely distributed to general neurologists and primary care physicians. In this case, the nomograms could provide an important service if they indicate a high likelihood of freedom from seizures after surgery, especially if the patient and doctor had not previously considered surgery. However, a major danger of such nomograms would be the implication that patients who are judged to have a low chance of becoming seizure-free should not be referred to an epilepsy centre. Nomograms should not replace the ability of the expert epileptologist, and the detailed presurgical assessment, to make a more informed prediction of postoperative seizure outcome. Use of nomograms by general neurologists, therefore, could do a great disservice if the result is to prevent the referral of these patients to specialised epilepsy centres, not only because the patients might still be excellent candidates for surgery, but also because, even if they are not candidates for surgery, they could benefi t from other treatments and approaches to psychiatric, psychological, and social management. These nomograms require prospective validation and refi nement before use, but if future versions are to be distributed to non-epilepsy specialists, they should Who will use epilepsy surgery nomograms, and why?