Abstract
Blumcke I, Spreafico R, Haaker G, Coras R, Kobow K, Bien CG, Pfäfflin M, Elger CE, Widman G, Schramm J, Becker A, Braun KP, Leijten F, Baayen JC, Aronica E, Chassoux F, Hamer HM, Stefan H, Rössler K, Thom M, Walker MC, Sisodiya SM, Duncan JS, McEvoy AW, Pieper T, Holthausen H, Kudernatsch M, Meencke HJ, Kahane P, Schulze-Bonhage A, Zentner J, Heiland DH, Urbach H, Steinhoff BJ, Bast T, Tassi L, Russo MG, Özkara C, Oz B, Krsek P, Vogelgesang S, Runge U, Lerche H, Weber YG, Honavar M, Pimentel J, Arzimanoglou A, Ulate-Campos A, Noachtar S, Hartl E, Schijns O, Guerrini R, Barba C, Jacques TS, Cross JH, Feucht M, Mühlebner A, Grunwald T, Trinka E, Winkler PA, Gil-Nagel A, Toledano Delgado R, Mayer TE, Lutz MB, Zountsas B, Garganis K, Rosenow F, Hermsen MAJA, Von Oertzen TJ, Diepgen TL, Avanzini G; EEBB Consortium. N Engl J Med 2017;377:1648–1656.
BACKGROUND: Detailed neuropathological information on the structural brain lesions underlying seizures is valuable for understanding drug-resistant focal epilepsy. METHODS: We report the diagnoses made on the basis of resected brain specimens from 9523 patients who underwent epilepsy surgery for drug-resistant seizures in 36 centers from 12 European countries over 25 years. Histopathological diagnoses were determined through examination of the specimens in local hospitals (41%) or at the German Neuropathology Reference Center for Epilepsy Surgery (59%). RESULTS: The onset of seizures occurred before 18 years of age in 75.9% of patients overall, and 72.5% of the patients underwent surgery as adults. The mean duration of epilepsy before surgical resection was 20.1 years among adults and 5.3 years among children. The temporal lobe was involved in 71.9% of operations. There were 36 histopathological diagnoses in seven major disease categories. The most common categories were hippocampal sclerosis, found in 36.4% of the patients (88.7% of cases were in adults), tumors (mainly ganglioglioma) in 23.6%, and malformations of cortical development in 19.8% (focal cortical dysplasia was the most common type, 52.7% of cases of which were in children). No histopathological diagnosis could be established for 7.7% of the patients. CONCLUSIONS: In patients with drug-resistant focal epilepsy requiring surgery, hippocampal sclerosis was the most common histopathological diagnosis among adults, and focal cortical dysplasia was the most common diagnosis among children. Tumors were the second most common lesion in both groups.
Commentary
In a recent edition of The New England Journal of Medicine, Blumcke et al. present the results of the European Epilepsy Brain Bank (EEBB) consisting of the largest, by far, case series of surgical epilepsy histopathology collected from 36 European centers over a span of 25 years. Their 9,523 cases were divided into seven categories (hippocampal sclerosis [HS], tumor, malformation of cortical development, glial scar, vascular malformation, encephalitis, and no lesion). The importance of large numbers needs little explanation; it will drive down the variation of point estimates to negligible figures and give rare cases a nontrivial platform. As the authors note, the EEBB collaborative effort led to the formation of task forces that developed the ILAE classification system for a histopathological classification system for HS and focal cortical dysplasias. Advances in epilepsy continue to be driven by large collaborations (1).
This cohort has some inherent limitations. The data was collected from a very large number of centers across multiple nations. While this may confer a certain amount of external validity, it raises concern regarding variability in the collection process. One presumes that samples were collected in all consecutive patients at all centers. Surgical skill, experience, and technique would almost certainly contribute some variability; some surgeons produce beautiful en-bloc resections while others are more liberal with the suction. These are all concerns that would naturally arise from any study involving so many contributing centers, although in this study, one does wonder whether a few centers dominate the dataset. In terms of accompanying clinical data, exactly what was collected is not completely clear in this article (a “prespecified minimal data set of deidentified clinicopathological information”) although it is reported to be nearly completely collected. The paper does not present any data on the critical patient outcomes such as seizure control, surgical complications or side effects, and so on, all of which were presumably collected. We welcome the publication of these findings in the near future.
More significantly, it appears that although the information was centrally curated, the actual pathological diagnoses were made locally, and no data is given regarding any attempt to determine interrater reliability. Though it is noted that the diagnoses were made by experienced neuropathologists, there is undoubtedly some variability in interpretation, particularly with the rarer cases. Aside from the aforementioned classification system for HS and focal cortical dysplasia, there do not seem to be agreed-upon standards or method for certifying competency. The authors note that the diagnostic yields were reassuringly similar between the German centers (59% of cases) and non-German centers (41%). Nonetheless, any data on variability of the interpretations would have been informative in their own right. The assessment of interobserver agreement was indeed performed earlier for a subset of patients in developing a classification system for HS by the same group (2). Similar analysis would have been equally welcomed on this high impact published dataset.
Certainly, their results are not unexpected; their findings validate smaller studies (3); furthermore, a little more than half of this dataset has been published in the past (4). Not surprisingly, HS is the most common finding, followed by low-grade tumor and malformation of cortical development. Oddly, one of the most interesting subgroups of patients presented in the earlier paper—dual pathology (comprising 5.2% of cases)—is not featured at all this time around. This is a rather unfortunate omission, as this distinction was clearly made by the authors internally. Perhaps it was an editorial decision to eliminate this category. If the information is available, it would be of particular value, even if it is an underestimate.
In some ways, this dataset describes the past generation of surgical epilepsy pathology, but it may not necessarily be as helpful a guide prospectively. For reasons that are still unclear, the epidemiology of surgical epilepsy has changed over the past 25 years (5, 6). There are fewer surgeries performed for HS, a greater number for nonlesional epilepsies, a smaller yield of surgeries per patient undergoing surgical evaluation, and possibly fewer surgeries overall. HS was found in 36.4% of specimens, which may become an answer to a multiple-choice epilepsy board question, but it no longer reflects what is experienced today, where seeing a patient with all the features of typical HS sometimes seems like an anachronism. Various hypotheses for this epidemiologic shift have been proposed, ranging from a clearance of backlog of more easily surgically amenable HS, a decreasing rate of new cases of HS, and the proliferation of smaller centers performing surgery for HS. Particularly considering data that the population of patients undergoing invasive intracranial monitoring without undergoing surgery is increasing (6), it is important to keep in mind that this is a collection of patients with medically refractory epilepsy who were offered and accepted resective surgery and whose specimens were in sufficiently good condition to undergo histopathological analysis. Thus, they represent a potentially shrinking fraction of all patients with medically intractable epilepsy. As this dataset spans this epidemiological change, it likely will provide further insight into this phenomenon in the coming years.
What can one expect from a similar biobank in the next 25 years? Ultimately, one hopes that the growth of this collection will slow; perhaps with better treatments, there may be fewer patients requiring ablative surgery for seizure control. This does not seem likely with medications alone, as recent evidence suggests that despite the proliferation of many new antiepileptic drugs, the number of patients remaining medically refractory has not changed during the past 30 years (7). However, a host of other treatments, such as neurostimulation, may eventually decrease the number of resective surgeries. Even if ablative surgery is performed, procedures that are safer (although without examinable pathology) may become more frequent, as laser ablation has already been reported to be used as first line for lesional temporal lobe epilepsy in some centers (8). Other even less-invasive ablative methods, such as focused ultrasound, have shown promise. Molecular and genetic pathology has redefined neuropathology of cancer and should also take bold steps forward in epilepsy neuropathology. Genetic information regarding both somatic and germline mutations will play a larger role in determining the likelihood of successful resective surgery, as certain mutations such as SCN1A are associated with poor outcome, whereas others are still surgically amenable (9, 10).
Despite these issues, what stays on my mind is that 10,000 is truly an astoundingly large number. Even at the large-volume epilepsy centers, the rate of surgeries is a trickle, never a flood. Procuring such a collection of surgical epilepsy histopathological cases takes patience and the ability to organize and collaborate across a large number of disparate centers. All these characteristics will be needed for the next 10,000 cases.