Humankind has long recognized epilepsy as a specific disease of the brain and tried to find remedies to treat it. Surgery for the epileptic brain, i.e., the removal of diseased or malformed cortex, began in the early stage of modern neurosurgery. Sir Victor Horsley, a brilliant pioneer of neurosurgery, performed operations on three patients with focal Jacksonian epilepsy in the 1880s [24]. In the early 20th century, brain mapping was widely studied and applied to epilepsy surgery as intraoperative cortical stimulation and mapping [4]. Montreal Neurological Institute (MNI) founded by Wilder Penfield became the center of these bold studies and practices. The invention of electroencephalography (EEG) by Hans Berger in 1924 greatly contributed to the understanding of epilepsy and heralded a new era of the objective diagnosis of epilepsy [32]. Through the decades that followed, EEG frequently combined with video recording of seizure semiology became part of standard presurgical evaluation [20]. In the latter half of the 20th century, temporal lobe epilepsy (TLE) became the main subject of research on epilepsy [4]. The physiology and function of the amygdala and hippocampus were extensively studied, which culminated in Scoville and Milner’s seminal paper on patients who received bilateral temporal lobectomies (including the famous patient, H.M.) [26]. This study revealed that the hippocampus is mainly involved in the retention of new memories, and subsequent studies literally revolutionized modern neuroscience. The diagnosis of TLE was firmly established, especially with the advent of computed tomography (CT) and magnetic resonance imaging (MRI). The greater efficacy of epilepsy surgery for TLE over medical therapy was clearly demonstrated in a randomized controlled trial published in 2001 [30]. In the surgical group, 58% of the patients were seizure-free at one year, whereas only 8% of the patients in the medical group were free of seizures. During the late 20th century, high-resolution MRI became the mainstay of diagnosis for epilepsy. The addition of other imaging modalities using nuclear medicine tools and merging various data to define an epileptogenic zone became the routine of many epileptologists. Understanding brain cortical malformation, especially focal cortical dysplasia (FCD), has changed greatly, and surgical resection of lesions transformed into a more precise and safe procedure. The advent and approval of vagus nerve stimulation (VNS) in the 1990s opened a new avenue toward neuromodulation therapy in epilepsy [33]. Responsive neurostimulation (RNS) has been introduced into clinical application since 2013 with promising results [7,8]. Currently, the use of RNS is off-label in children with intractable epilepsy [12].

This is a brief history of epilepsy surgery since the late 19th centuries. In 2019, what we are facing in the field of epilepsy surgery is a rapid and fundamental change in what we once thought usual and standard. In 2017, the International League Against Epilepsy (ILAE) introduced a revised system of classification for seizure types and epilepsies [5,23]. The differentiation of seizure types is the basic starting point for patient care and an initial diagnosis. Some seizure types are strongly associated with the diagnosis of specific epilepsy syndromes and treatment options. Many anti-seizure drugs (ASDs) are most efficacious for certain types of seizures. Some surgical procedures are also particularly effective for certain types of seizures, for example, callosotomy for drop attacks. Finally, the world-wide communication and sharing of common nomenclatures for various phenomena related to epilepsy is mandatory. The previous system of classification for seizure types and epilepsies was introduced by the ILAE in the 1980s [34,35]. There have been few changes in this scheme for more than 30 years, although many argued the needs for reappraisal [3]. Scientific progress, new concepts about epilepsy, and even changing social interest in epilepsy urged the reappraisal of the classification scheme. One of the major changes in the 2017 revision is to replace the term ‘partial seizure’ with ‘focal seizure’. Although the meaning (originating from only one hemisphere) is almost the same, ‘focal’ has a more anatomical connotation than the word ‘partial’ [5]. Likewise, the long-used terminology ‘complex partial seizure’ was substituted with ‘focal impaired awareness seizure’ for clarity for both medical personnel and laypeople. For the classification of epilepsy, multilevel diagnostic steps from seizure type to epilepsy type and further to epilepsy syndrome have been proposed. Etiological considerations at every step is emphasized and reflect the enormous amount of development in the etiological diagnosis of epilepsy, especially in structural and genetic causes. Changing long-held jargon is a hard task for many clinicians, and changing concepts is harder than just stating them. It is hoped that the revised classifications may enhance the understanding of epilepsy and encourage the anatomic and surgical imaginations of many surgeons.

Paradigms in drug discovery are changing and many novel ASDs have emerged for clinical use. The role of epilepsy surgery may change according to this trend. Since phenobarbital was introduced in 1912, many efforts have been made to develop effective ASDs against epilepsy. In the 1930s, through screening via an electroshock animal model, phenytoin was discovered as an effective ASD. Thereafter, many compounds were validated with screening tools. However, there was a period of stagnation between the early 1960s and mid-1970s, when only two major ASDs (valproate and carbamazepine) were adopted [21]. In 1975, U.S. National Institutes of Health (NIH)/National Institute of Neurological Disorders and Stroke (NINDS) launched the Anticonvulsant Screening Program (ASP), a systematic screening program for potentially useful compounds. The ASP led to the discovery of many socalled third generation ASDs such as vigabatrin, lamotrigine, topiramate, and levetiracetam [17]. Behind this success, an important criticism has been made against the ASD development process [27]. Even novel ASDs are largely ineffective in drug-resistant epilepsy patients who comprise 20-30% of all patients. This means that current screening tools cannot sufficiently rule out less effective candidates or select compounds potentially effective in certain types of seizures. A unique example is levetiracetam. Levetiracetam initially failed to pass the screening tests using acute seizure animal models. However, levetiracetam showed robust suppression of seizure development in chronic epileptic animal models [11]. In 2015, the ASP was reformulated as the Epilepsy Therapy Screening Program to incorporate drugs with antiepileptic properties [17]. To date, the process of ASD development has evolved from the use of serendipity to systematic screening. Recently, molecular genetic studies have found many novel targets of epilepsy against which specific drugs can be developed. Everolimus, an mTOR inhibitor, is used to control epilepsy in tuberous sclerosis complex (TSC) [13]. Interestingly, sequencing studies discovered that hemimegalencephaly is caused by somatic mutations of PI3KAKT3-mTOR pathway [14]. FCD was also found to arise from somatic mutations of the TSC-mTOR pathways, raising the possibility of applying mTOR inhibitors to a wide range of epilepsy patients [15,16]. Therefore, for patients with major malformations such as hemimegalencephaly that require extensive and challenging operations, epilepsy surgery may concede to tissue biopsies for sequencing studies and molecular targeted agents can be used for definite treatment.

For nearly a century, the core aim of epilepsy surgery has been the extirpation of diseased/malformed cortex. The identification of the brain areas responsible for seizure generation (epileptogenic zone) using every means and tool became the central tenet of neurology [22]. However, the concept of the epileptogenic zone was criticized as much as it was pursued. Above all, epileptogenic zone can be identified only after surgical removal, not prior to the operation which is more desirable. Defining the epileptogenic zone with all available resources cannot completely prevent surgical failure. Furthermore, some patients experience seizure recurrence years after surgery with an interval of seizure freedom. This phenomenon is not compatible with the concept of the epileptogenic zone. The common practice of prescribing ASDs to surgical patients for at least a few months after surgery seems to be a compromise between the concept and reality. Recently, accumulating evidence has indicated that focal epilepsy stems from multiple structures that comprise neuronal circuits [28]. Multiple regions of abnormal circuits can act as independent nodes in a pathological network, underlying surgical failure if only a part of those regions are destroyed. In this view, epilepsy is a network disorder, and focal epilepsy is not focal, as we thought. Even when a highly epileptogenic lesion, such as hypothalamic hamartoma, exists, the neocortical region can be an independent source of seizures [25]. That many patients with epilepsy have comorbid neuropsychiatric disorders supports the existence of brain network abnormalities in epilepsy [28]. Epileptic patients have a 43% higher risk of depression than normal population [6]. Obsessive-compulsive disorder is also highly associated with chronic epilepsy, notably TLE [10]. Much of the evidence for the network hypothesis was obtained from studies using depth electrodes and SEEG. The network hypothesis again in turn strongly supports the use of SEEG for the presurgical evaluation of epilepsy patients. If seizures originate from a widespread network including deep subcortical structures, mapping with SEEG is superior to an invasive study using subdural grids. Pioneered by Jean Talairach and Jean Bancaud in the 1950s, SEEG has gained worldwide popularity [20]. SEEG is less invasive and safer than subdural grids, which have higher complication profiles [29]. SEEG is more suited for the evaluation of deep-seated areas, multilobar/bilateral exploration, and the mapping of functional networks [18]. The SEEG procedure can be automated to frameless robot-assisted positioning of electrodes [19]. If a malfunctioning network is to blame in epilepsy, it is possible to destroy or isolate epileptogenic nodes in the network to quench a seizure rather than to shatter the entire network. SEEG electrodes installed for diagnosis can be co-opted for therapeutic intervention using radiofrequency thermocoagulation or laser interstitial thermal therapy [1,2]. The emerging neuromodulatory therapies for epilepsy such as deep brain stimulation (DBS), VNS, and RNS, owe much of their efficacy to the network-like nature of epilepsy in the generation and propagation of seizures, although much has to be discovered about their exact mechanism [9,31].

Epilepsy surgery has developed for more than one hundred years. In 2019, we are facing great changes in the paradigms of epilepsy diagnosis, medical treatment, and surgery. The emerging concept of epilepsy as a network disorder is the core of these many changes. What we should remember is that too many patients with intractable epilepsy remain. We need greater consideration of past footprints and discarded pathways to discover novel therapies for this fatiguing disease.