1 Epileptic spasms (ES) are a unique type of seizure [1] that classically occur in infants, as infantile spasms, with a certain age dependence. Symptoms may also develop after infancy or return later after an initial remission of infantile spasms. ES have been reported in many conditions, including malformation of cortical development (MCD), brain injury, perinatal asphyxia, and genetic abnormalities. Generally, patients with ES develop other epileptic symptoms, such as generalized tonic-clonic seizures or various types of focal seizures, by the age of three to five [2, 3]. Data on the surgical outcomes of children with ES, however, are scarce. ES are often considered to be a factor associated with poor prognosis after surgery; more importantly, the electroclinical characteristics of ES and their influence on postoperative efficacy are still unclear. The localization of the epileptogenic zone is difficult due to the symmetry of symptomatology and the extensiveness of the seizure onset, which makes epilepsy surgery for ES challenging. However, a recent paper indicated that not only is surgery a viable option for appropriately selected ES patients, but that postoperative efficacy is also comparable to that achieved for other types of focal epilepsy [4]. Considering that ES may be a potential type of focal epilepsy, focal resection is promising; besides, surgical treatment can improve neurocognition while curing epilepsy [5]. Advanced neuroimaging and invasive monitoring techniques have helped us to locate epileptogenic foci, so that patients of this type, who were not previously recommended for surgery, can become seizure-free upon resection of the foci. In an article from the United States, of 65 infants and children with drug-resistant ES who were treated with surgery, those patients with early surgical treatment or consistency between MRI lesions and EEG abnormalities tended to obtain better surgical outcomes [6]. Even if there is no specific lesion on MRI, a good surgical outcome can be achieved when there are focal metabolic abnormalities on PET. According to an Italian study, completeness of resection of the seizure onset zone and of the identifiable lesion is highly associated with a better surgical outcome [7-9]. Here, we retrospectively analysed 64 children with drug-resistant spasms, who we had operated on in Beijing, in order to discuss the aetiology, clinical features, operation procedures, and prognosis in detail. Our purpose was to explore the relationship between surgical efficacy and these factors, as well as to propose an appropriate preoperative scheme for drug-resistant ES.

Methods

Results

Surgery

11 For 61 (95%) patients, intraoperative electrocorticography (ECoG) monitoring was performed. Twenty-nine (45%) patients underwent left-side surgery, and 35 (55%) underwent right-side surgery. Thirteen (20%) patients received SEEG due to non-localizing data, and 11 of them had favourable outcome. There was no significant correlation between surgical result and whether or not SEEG was used. We performed 22 (34%) unlimber resections, 17 (27%) hemispherotomies (total/subtotal), 15 (23%) tuberectomies, and seven (11%) multilobar resections; the remaining three cases received RF-TC treatment. Among the 64 patients, the frontal lobe was operated on in 16 (25%), 12 of whom had favourable surgical results; multiple lobes were operated on in 13 (20%) patients, 12 of whom had favourable results; and 13 (20%) patients underwent hemispheric surgery, 12 of whom had favourable results. In addition, the parietal lobe, temporal lobe, and occipital lobe were operated on in 11, eight, and three patients, respectively. There were no significant differences in surgical type or resective site between the favourable and unfavourable groups (figure 1).

Figure 1

Data on pathology (A), type of surgery (B) and site of resection (C) in patients with favourable or unfavourable outcome.

Discussion

16 Many studies have indicated that ES may potentially be a focal-type seizure, and some have reported similar postsurgical results between focal ES and other focal seizures, therefore surgery may provide these patients with an opportunity to become seizure-free [13, 14]. The features of focal seizures include a focal lesion, focal discharges or onset, focal hypometabolism on PET, and asymmetry of symptomatology; surgical results may be highly correlated with these features. In addition, surgical outcome may also be affected by aetiology, surgical type, and surgical site. Few studies have examined the influence of these factors on the prognosis of ES [15, 16]. The present case series from Beijing, involving 64 children with refractory ES, provides a deeper understanding of these factors and may help guide future surgical treatment.

17 Approximately 83% of the patients in our study had favourable surgical outcomes, a much higher proportion than we expected and also higher than the average success rates in recent studies from other centres [17, 18]. We did not find any differences in sex, age at first seizure, age at surgery, or duration of seizures between the groups with favourable and unfavourable surgical outcome. Although children with intractable ES who had surgery within three years of their first seizure tended to have better outcome based on a previous study [19], we did not find such a result in our study. However, according to our data, the seizure onset age in 53 (82.8%) cases was less than one year of age, and the duration between seizure onset and surgery in 48 (75%) cases was less than three years, which may account for the good surgical results. The reason for achieving better outcomes at a younger age is not clear. A possible explanation is that ES are more active in young children. It is easier to establish a large epileptic network and form secondary epileptic foci in infants and young children. Surgical treatment should be performed as soon as possible, which can provide early relief and promote the resumption of brain developmental progression. Meanwhile, larger resections in these patients can reduce injury due to functional compensation at a later stage [6]. Some studies have suggested that symmetric spasms or the coexistence of symmetric and asymmetric spasms are adverse factors for long-term prognosis, as they are usually associated with extensive lesions. Whether asymmetrical symptomatology suggests a better prognosis is unclear. In our study, we did not find a correlation between surgical outcome and asymmetrical symptomatology or a combination of other types of seizure. This suggests that symptomatology is not very reliable in indicating prognosis, but asymmetry in the intensity of jerking or face twitching during spasms does help us to lateralize the hemisphere responsible for producing ES [20]. In the present study, 53 (83%) children with ES had a definite lesion on MRI, which also accounts for the good surgical results. It has been reported that about half of children with ES have identifiable lesions on MRI [6].

18 Although postoperative results are not associated with any specific imaging features, the concordance between imaging and EEG has important clinical implications. VEEG plays an important role in determining surgical prognosis. Studies have suggested that if hypometabolism on PET is consistent with ictal or interictal abnormalities on EEG, resection of that region would result in seizure cessation and improved cognitive behaviour [21, 22]. In our study, there was very good correlation between surgical outcome and concordance between MRI and IEDs and concordance between PET and IEDs. The results of multivariate regression analysis in the present study also support this finding, which further confirm the importance of combining imaging and EEG data in preoperative evaluation. However, we should be cautious to not exclude patients who have negative MRI or EEG presenting non-localized discharges, especially in children. Children with epileptic spasms and diffuse epileptic discharges, that have been shown to be compatible with focal lesions, may achieve good surgical outcome after surgical resection. Furthermore, the development of imaging and EEG postprocessing techniques may provide new focal evidence. Additionally, interictal high γ frequency was found to be correlated with favourable surgical outcome. Asano et al. showed that spasms are associated with either a “leading” spike followed by fast bursts or fast bursts without a “leading” spike [23, 24]. Recent studies have shown detectable focal HFOs, especially in the high γ band (around 60-90 Hz) on scalp EEG recordings in children with epileptic spasms. Asymmetric ictal γ activity in the centroparietal area is associated with observable asymmetry of the spasms. Thus, resection of cortical areas, represented by high γ frequency, highly correlated with good surgical outcomes [25, 26]. In our study, we found no significant correlation between surgical outcome and surgical type or resection site. The most commonly used surgical procedures in our patients were unilobar resection, hemispherotomy (total/subtotal), and tuberectomy, demonstrating that ES are initiated as cortical epileptic discharges which, during a “critical” developmental period, undergo secondary generalization in an age-dependent mechanism to emerge as spasms, despite the complex mechanisms and brain network associated with ES [27]. The focal cortical origin of spasms has also been demonstrated by invasive EEG recordings as well as favourable outcomes following removal of discrete epileptogenic lesions [6, 22, 28]. No studies have shown a correlation between surgical site and surgical outcome. The most common surgical site in our study was the frontal lobe, even after SEEG. This is consistent with a previous finding showing that these complex clusters of axial attacks are more likely to form in the frontal lobes; the exact mechanism requires further study of brain networks [29]. We found no significant correlation between surgical outcome and whether or not SEEG was performed, suggesting that there are limitations in the application of this technique for ES.

19 By grouping the patients with no seizure after surgery, we can provide a clinical basis for future surgical strategies. First, whether ES patients with negative MRI can undergo surgical treatment and the extent of surgical resection have always been concerns of surgeons. Our data indicate that in MRI-positive cases, due to extensive cortical injury, such as generalized cortical dysplasia, TSC, or perinatal hypoxia, being a common cause of spasm seizures, localized resection is relatively rare. The surgical decision mainly depends on the preoperative EEG and imaging data, especially low focal metabolism on PET and its concordance with interictal discharges, therefore the range for resection is generally wide. In MRI-negative cases, SEEG is performed at a higher rate, leading to a corresponding increase in limited and tailored resection, which also accounts for a favourable outcome. Therefore, if accurate resection and functional protection are considered, the SEEG is still a necessary monitoring tool. In addition, we found that the MRI-negative group was more dependent on intraoperative ECoG monitoring, which was also one of the main indexes for surgical resection range and surgical result. However, in the MRI-positive group, due to the extensive lesions and dispersed discharge, intraoperative ECoG monitoring had some limitations. For the two most common causes of ES, FCD and TSC, we found that concordance between MRI, PET, and scalp EEG was an important factor in determining surgical strategy. Therefore, preoperative high-resolution MRI, precise interpretation of EEG, and fusion technologies of PET and MRI all provide an important basis for surgical decision-making. In addition, the proportion of SEEG application in the TSC group was much higher than that in the FCD group, suggesting that although TSC patients have clear lesions on imaging, intracranial electrodes are often needed to help identify the true epileptic nodules due to the multiple lesions. As imaging techniques have improved, FCD has become more recognized, and SEEG is performed less. However, the proportion of patients with limited resection in the FCD group was lower than that in the TSC group, perhaps because the boundaries of FCD were sometimes relatively unclear; the extent of resection was often beyond our expectation, involving the corresponding cortical or subcortical structures. According to our data, the proportion of patients with continuous discharges on ECoG was relatively high in the FCD group, therefore the intraoperative ECoG could be an important index for the surgical resection range for FCD. Whether intracranial electrodes are needed for ES is of more concern to surgeons and patients because the onset of ES usually involves a widespread slow wave, and the ictal discharges are also diffuse even after electrode implantation. We found that with less valuable information based on preoperative evaluation, such as focal lesions, focal interictal discharges, and focal hypometabolism, SEEG tended to be performed in order to identify epileptogenic foci, consistent with previous studies [30]. In our study, although SEEG was not the main factor in determining the prognosis, the implantation of intracranial electrodes narrowed the range of resection, which was more consistent with the concept of accurate resection, particularly beneficial regarding functional protection of older children.

20 Based on the above factors, a workflow for presurgical evaluation in children with epileptic spasms is proposed (figure 2). We hope that the preoperative evaluation process for children with ES can be simplified and a reasonable surgical plan can be made accordingly.

Figure 2

Proposed workflow for presurgical evaluation in children with epileptic spasms.

Conclusion

22 Our study shows that surgical resection can be effective for children with drug-resistant ES when certain indications are met. Concordance between MRI and IEDs, concordance between PET and IEDs, interictal high γ frequency, and continuous discharges on ECoG are important factors associated with a favourable surgical outcome. We propose a workflow for presurgical evaluation in children with ES to simplify the process and help provide a reasonable surgical plan.