Early onset medically intractable epileptic encephalopathies are often caused by hemispheric pathology such as malformations of cortical development (MCDs), perinatal stroke, Sturge-Weber syndrome, or Rasmussen encephalitis.
Until the late 1960s, complete removal of the affected hemisphere (i.e., anatomic hemispherectomy) was the preferred surgical treatment modality for these patients, and approximately 70–80% of them have been reported to become seizure-free. However, early and delayed surgical complications including significant blood loss, coagulopathy, metabolic imbalances, superficial cerebral hemosiderosis, and hydrocephalus with mortality rates up to 30% lead to a significant decline in use (Krynauw, 1950; Oppenheimer & Griffith, 1966; Rasmussen, 1973; Vining et al., 1997; Bahuleyan et al., 2012).
Following the concept of functional hemispherectomy introduced by Rasmussen, various disconnection techniques have been developed during the last two decades. The common goal of these techniques is to avoid/reduce the above-mentioned complications without minimizing postoperative seizure outcome (Rasmussen, 1983; Delalande et al., 1992, 2007; Schramm et al., 1995; Villemure & Mascott, 1995; Carson et al., 1996; Peacock et al., 1996; Vining et al., 1997; Shimizu & Maehara, 2000; Cook et al., 2004; Villemure & Daniel, 2006; Limbrick et al., 2009; Bahuleyan et al., 2012; Schramm et al., 2012).
In practice, two different approaches for hemispheric disconnection are currently used: lateral periinsular hemispherotomy with various modifications that has consistently been reported to be safe and effective and vertical parasagittal hemispherotomy introduced by Delalande et al. (Villemure & Daniel, 2006; Delalande et al., 2007; Schramm et al., 2012).
Compared with the lateral techniques, the vertical approach seems to have the following advantages: It allows the complete disconnection of the hemisphere at the level of the thalamus, obviating the need to open and dissect the subarachnoid space of the Sylvian fissure and to resect the insula.
However, apart from Delalande′s report published in 2007, no studies from other centers replicating his results and/or reporting long-term experience with this technique have been reported (Chandra et al., 2008; Kawai et al., 2013).
At the Vienna pediatric epilepsy center, vertical parasagittal hemispherotomy has been performed since 1998. We report on our experience in 40 children and adolescents with special emphasis on long-term efficacy and safety.
- Top of page
We report our experience with a single disconnective technique. Results from single-center studies comparing different techniques by using historical controls support the evolution toward less resective and more disconnective techniques (Cook et al., 2004; Kwan et al., 2010).
Vertical perithalamic hemispherotomy was implemented at our center in 1998. After thorough study of the techniques described at that time including the historical French experience with regard to the disconnection plane at the level of the basal ganglia, we adopted the technique developed by Delalande (Laine & Gros, 1956; Rasmussen, 1973; Delalande et al., 1992; Schramm et al., 1995; Villemure & Mascott, 1995; Kanev et al., 1997). We thought the perithalamic incision would help to disconnect any dysplastic gray matter between the cortex and the ventricular wall. Insular cortex and any adjacent dysplasia are disconnected with no need to dissect between the middle cerebral artery (MCA) branches. In addition, the main axis of the surgical approach being sagittal should reduce the risk of inadvertently damage to the contralateral hemisphere. Our experience led us to use this technique independent of the underlying pathology.
Anatomic hemispherectomy is rarely used today owing to well-known early and delayed surgical complications (Krynauw, 1950; Oppenheimer & Griffith, 1966; Rasmussen, 1973; Vining et al., 1997; Bahuleyan et al., 2012). Several hemispherotomy techniques have been developed to minimize the extent of resection while maintaining the same seizure outcome.
We think, that—compared with vertical parasagittal hemispherotomy—all lateral techniques have in common several drawbacks:
First, a considerable although various amount of the brain has still to be removed, thus increasing the risk of intraoperative blood loss (sometimes associated with hypovolemia and death; Brian et al., 1990). Although—compared with anatomic hemispherectomy—lateral hemispherotomy techniques led to a significant decline in blood loss and requirement for blood transfusion, the reported number of patients needing blood replacement is still 10–50% (Limbrick et al., 2009; Scavarda et al., 2009; Schramm et al., 2012). In our series, only two very young children (5.0%) needed blood replacement. Similar to the vertical technique the transsylvian keyhole technique described by Schramm uses only very limited resection, but its use seems to be restricted to patients with postischemic porencephalic defects.
Second, the subarachnoid space of the Sylvian fissure needs to be opened and dissected. The major risk of functional hemispherectomy/hemispherotomy with minimal tissue resection seems to be swelling related to vascular compromise. We think that avoiding any manipulation of the insular or opercular MCA branches minimizes the risk of vasospasm and related space-occupying ischemia, especially in patients apart from those with porencephalic cysts. In addition, a wide opening and dissection of the Sylvian fissure might contribute to the development of hydrocephalus. The vertical route of disconnection has the advantage that access to the lateral ventricle is accomplished via a small corticotomy tunnel obviating the need to open and dissect the subarachnoid space of the Sylvian fissure. This is suggested to minimize the risk to develop hydrocephalus (Lew et al., 2010). Even though lateral hemispherotomy techniques have successfully reduced the hydrocephalus rate encountered in hemispherectomy (Vining et al., 1997; Shimizu & Maehara, 2000; Devlin et al., 2003; Cook et al., 2004; Villemure & Daniel, 2006; Limbrick et al., 2009; Kwan et al., 2010; Schramm et al., 2012), a recent multicenter analysis performed by the Posthemispherectomy Hydrocephalus Workgroup still found an overall hydrocephalus rate of 20% in patients operated with various lateral hemispherotomy techniques (Lew et al., 2010). However, other large series using lateral hemispherotomy techniques reported hydrocephalus rates between 2% and 20%, indicating that the individual hemispherotomy technique and/or other surgical nuances also influence the risk to develop hydrocephalus (Shimizu & Maehara, 2000; Villemure & Daniel, 2006; Cats et al., 2007; Schramm et al., 2012).
One of these surgical nuances that possibly explains different extents of impairment of CSF may be the contamination of the CSF with blood products and tissue debris. To overcome this problem of the contaminated CSF, some authors routinely leave a ventricular catheter for some days to drain the CSF (Villemure & Mascott, 1995). However, this procedure carries the potential risk of overdrainage associated with venous hemorrhage occurring topographically at a distance from the surgical site, for example, in the unaffected hemisphere (deRibaupierre et al., 2004). At our center, we extensively irrigate the ventricular system with synthetic liquor after completion of the disconnection until the CSF is perfectly clean. With this protocol, only one patient (2.5%) needed shunt placement.
In contrast to our findings, Delalande reported a shunt rate of 15.7%. Possible explanations for the better result in our patients may be the above-mentioned extensive irrigation of the ventricular system, and the closure of the corticotomy tunnel with fibrin flue and Lyostypt, which helped to restore a normal CSF circulation. This assumption is supported by reports from other centers, where postoperative subdural hygroma formation was reported to be avoided by closure of the corticotomy with fibrin glue (Blauwblomme & Harkness, 2010). Another factor might be the higher rate of MCD patients in the series of Delalande (36%) compared with our series (27.5%).
Finally, the insular cortex and/or its subcortical structures remain in place or have to be resected in lateral techniques (Cook et al., 2004; Villemure & Daniel, 2006; Schramm et al., 2012). The role of the insula in epilepsy has received increased interest over the last decades, and several studies demonstrated ictal onset within the insula mimicking frontal, temporal, and parietal epilepsy (Nguyen et al., 2009). Insular contribution was also demonstrated in hemispheric epilepsy. Recent reports from hemispherotomy series evaluating seizure outcome with respect to the extent of insular resection demonstrated that residual insular cortex was associated with lower rates of seizure-free patients (Cats et al., 2007). This could be particularly important in MCDs, as dysplastic neurons were also identified in the striatum subserving a potential contributor in epileptogenesis and seizure spread (Kaido et al., 2012), and increased rates of seizure control were found when the insular cortex and its subcortical structures were removed (Cook et al., 2004). This may in part be responsible for less favorable seizure outcomes reported in patients with MCDs after hemispherotomy (31–70%) as compared to 71–90% in patients with acquired pathologies (Devlin et al., 2003; Kossoff et al., 2003; Schramm et al., 2012). The vertical parasagittal hemispherotomy proposed by Delalande currently is the only technique that allows complete disconnection at the level of the thalamus. Thereby, the insular cortex as well as possible dysplastic epileptogenic neurons within subcortical structures can be disconnected parallel to the axis of the perforating vessels. In lateral approaches these structures need eventually to be resected (Cook et al., 2004). This perithalamic disconnection might be one explanation for the higher percentage of MCD patients with a favorable seizure outcome in the series of Delalande (75%) and our own series (91%).
In summary, the data from Delalande and our long-term follow-up data compare favorably with the published literature of lateral techniques (Cook et al., 2004; Di Rocco et al., 2006; Villemure & Daniel, 2006; Limbrick et al., 2009; Kwan et al., 2010; Schramm et al., 2012; Wiebe & Berg, 2013).
- Top of page
Our study is limited by a relatively small sample size. However, no children were lost to follow-up (as seen in many other studies). Even though we present long follow-up data for at least 1 year in 92.5%, 2 years in 75.0%, and 5 years and beyond in 27.5% of the patients, we have to state that outcome may be worse with longer follow-up.
However, data were collected prospectively on an inpatient basis and not via telephone interviews or via email questionnaires.
The aim of this study was to evaluate safety and efficacy with respect to seizure outcome of a specific neurosurgical technique. The relation between surgical and developmental outcome is more complex and multifactorial (i.e., also depending on age at seizure onset and duration of the disease before surgery, AEDs used, underlying pathology, and time of AED withdrawal). Therefore, detailed developmental data were not included in this analysis. However, developmental progress was noticed in all seizure-free patients, and there was no deterioration after surgery in any of our patients.