There have been only a few large series that have used a tailored temporal lobectomy.
To clarify whether tailoring a temporal lobe resection will lead to equivalent epilepsy outcomes or have the same predictive factors for success when compared with standard resections.
Retrospective analysis of 222 patients undergoing a tailored temporal lobe resection. Demographic measures and typical factors influencing outcome were evaluated.
Pathology included 222 cases. With a mean follow-up of 5.4 years, 70% of patients achieved Engel class I outcome. A significant factor predicting Engel class I outcome on multivariate analysis was lesional pathology (P = .04). Among patients with hippocampal sclerosis, extent of lateral neocortical resection and hippocampal resection were not statistically associated with Engel class I outcome (P = .93 and P = .24). However, an analysis of Engel class subgroups a to d showed that patients who had a complete hippocampectomy in the total series were more likely to achieve an Engel class Ia outcome (P = .04). This was also true among patients with hippocampal sclerosis (P = .03). Secondarily, generalized seizure (P = .01) predicted outcome less than Engel class I. Predictive of poor outcome was the need for preoperative electrodes (P = .02). Complications included superior quadrant visual field defects, 2 cases of permanent dysphasia, and 3 wound infections.
Predictors of successful seizure outcome for a tailored temporal lobectomy are similar to standard lobectomy. Patients with secondarily generalized epilepsy and cases in which preoperative subdural electrodes were thought necessary were less likely to achieve class I outcome. Among Engel class I cases, those who had a complete hippocampectomy were more likely to achieve Engel class Ia outcome.
Temporal lobe epilepsy constitutes nearly two-thirds of all cases, making it the most common focal epilepsy.1–4 Seizures often arise from mesial temporal structures including the amygdala, hippocampus, and parahippocampal gyrus.3,5 Medically intractable temporal lobe epilepsy is commonly associated with hippocampal sclerosis, which is the most common structural abnormality and histopathological finding in adult epilepsy.6–8 For surgical candidates undergoing anterior temporal lobectomy, seizure freedom is achieved in 60% to 80% of cases.9–14 However, temporal lobe epilepsy is not limited to hippocampal sclerosis and may include other pathologies such as tumor, heterotopias, and gliosis. It is not clear whether tailoring a temporal lobe resection for medically intractable epilepsy will lead to equivalent epilepsy outcomes or complication rates or have the same risk factors for success and failure in comparison with standard resections. Kanner et al15 reported that 87% Engel class I with a mean follow-up of 38.5 months and that the extent of mesial resection did not predict seizure outcome when the resection is tailored.
Kanner et al15 demonstrated that limited resection of medial temporal structures is not necessarily associated with a poor seizure outcome. Few series have been reported using the tailored corticoamygdalohippocampectomy approach.8,16,17 The technique of tailored temporal lobectomy allows for a choice of the extent of resection of the lateral temporal lobe and the hippocampus as dictated by all of the patient's pre-operative and intra-operative findings. Factors to consider include magnetic resonance imaging (MRI), electroencephalography (EEG), cortical mapping, severity of epilepsy, memory function, and intraoperative electrocorticography. The patient group with hippocampal sclerosis is a relatively homogeneous group that allows for a fair evaluation of the technique of tailored temporal lobectomy. To determine patient outcomes with the use of this approach and discover factors leading to best outcomes, a retrospective evaluation of our epilepsy surgery database was done.
PATIENTS AND METHODS
An institutional review board-approved retrospective review of our prospectively maintained institutional epilepsy surgery database was queried for cases of temporal lobectomy. There were a total of 233 cases in which 222 consecutive nonselected patients with at least 1 year follow-up (mean 5.4 years) were included in this study. Pathology included 126 cases of hippocampal sclerosis, 39 low-grade tumors, 4 heterotopias, 3 cavernomas, a posttraumatic scar, and 49 with nonlesional pathology or gliosis. Those 11 patients with missing data, less than 1 year follow-up, multilobar cases, or a different surgical approach were excluded. None of these excluded patients had major morbidity. All surgeries were performed at a single institution (Rush University) by a single surgeon (R.W.B.). All relevant demographic measures and factors known to affect epilepsy outcome were evaluated. The sample group consisted of 99 males and 123 females. Mean age was 33 years old. The primary end point is the analysis of factors leading to Engel class I outcome, whereas the secondary end point is the analysis of the effect of extent of resection on outcomes when using the tailored approach.
All data relevant to postoperative surgical outcome were evaluated by using univariate (χ2, Fisher exact test, t test, and Wilcoxon rank sum test) and multivariate analysis (logistic regression analysis) with attention to the effect of the extent of resection of the hippocampus as it relates to epilepsy outcome as measured by Engel classification. A complete hippocampal resection was considered a resection to the level of the tectum. A partial resection was taken to the level of the choroidal point. Engel classification was used to measure outcome at follow-up. Class I is divided into class Ia, Ib, Ic, and Id, with class Ia being completely seizure free.
Preoperative evaluation included case discussion of each patient at a multidisciplinary conference in which it was agreed that the patient was medically intractable and a surgical candidate. All patients had complex partial seizures, and the generalized category in our analysis was for those patients that had secondary generalized convulsive seizures. Patients underwent video EEG monitoring as well as epilepsy protocol MRI. Imaging in most cases included a 1.5 T MRI, and if indicated, a 3 T MRI for selected cases. Additional imaging, when indicated, included positron emission tomography, single-photon emission computed tomography, magnetoencephalography, functional MRI, and subtraction ictal single-photon emission computed tomography coregistered to MRI. Preoperative neuropsychological testing and evaluation was performed. Wada testing was performed as indicated. Intracranial electrodes, including subdural and depth electrodes, were first inserted and patients monitored for mapping in cases where there was unclear localization, in cases of possible bitemporal localization (subtemporal strips), and in lateral neocortical cases in the dominant hemisphere.
The surgical procedure consisted of a craniotomy under general anesthesia, electrocorticography (Figure), followed by a corticoamygdalohippocampectomy of varying dimensions ranging from 2 to 6 cm on the middle temporal gyrus (mean 3.5 cm, standard deviation 1.6 cm) as measured at the time of surgery. The extent of resection for each patient was the result of a careful analysis of preoperative factors, combined to a lesser degree with intraoperative electrocorticography. Factors that led to a decision to perform a smaller lateral cortical resection or a partial hippocampectomy included: excellent preoperative neuropsychological scores, proximity to speech cortex as determined with cortical stimulation, resolution of intraictal epileptic discharges after a partial hippocampal resection, and normal MRI without hippocampal sclerosis.
Superior temporal gyrus resections averaged 2 cm with a range of 0 to 8 cm (standard deviation 1.9 cm) with the superior temporal gyrus often being spared in the dominant hemisphere. Neocortical resection varied based on side, MRI findings, and preoperative and intraoperative EEG. The amygdala was resected to the level of M1 and the endorhinal sulcus. The hippocampus was resected to the level of the colliculus for a complete resection or to the level of the choroidal point for a partial resection. Complete resection was confirmed intraoperatively as a resection to the posterior medial level of the hippocampus around the posterior thalamus and exposure of the anterior portion of the atrium. This was confirmed on postoperative MRI imaging. Electrocorticography was repeated, and further resection was occasionally done. We estimate that electrocorticography findings of electrographic seizure (not temporal spiking, as is commonly noted) after initial resection led to further resection in approximately 5% of cases.
Table 1 lists the demographics of those patients evaluated. The sample group consisted of 99 males and 123 females. Mean age was 33 years. Table 2 lists the breakdown of patients based on pathology. With a mean follow-up of 5.4 years, 70% of patients achieved Engel class I outcome. There were a total of 156 of the 222 patients who achieved Engel class I outcomes of which 94 had hippocampal sclerosis (Table 3). Seventy-six percent of patients with lesional pathology had an Engels class I outcome, whereas patients with nonlesional pathology had a 54% chance of Engel class I outcome overall. This statistically significant difference was shown on both univariate (P = .001) and multivariate analysis (P = .04, odds ratio [OR] = 2.1, confidence interval [CI] = 1.03,4.31). In the multivariate analysis, independent variables were the extent of hippocampectomy, the extent of lateral cortical resection, length of follow-up, and all significant factors from univariate analysis. Demographic data and other factors were evaluated for their possible influence on Engel class I outcome. The only demographic factor that was found to be statistically significant in correlation with Engel class I outcome was an absence of a history of secondary generalization. This was true when considering all cases (P = .01 univariate only) and with the uniform group of hippocampal sclerosis cases alone (P = .004 univariate, P = .03 multivariate). Twenty-five patients (11%) had Engel class IV outcome. A need for preoperative intracranial electrode recording was the only factor associated with a higher likelihood of Engel class IV outcome (P = .01 on univariate and multivariate analysis OR = 3.53, CI = 1.28,9.73).
There was no statistical difference in likelihood of Engels class I outcome between patients who had a complete hippocampectomy (n = 113) and a partial hippocampectomy (n = 82) (P = .47). This remained true when controlling for pathology, extent of neocortical resection, and other factors including length of follow-up and preoperative intracranial electrode recording. Table 4 looks at the comparison among those in Engel class I. An analysis of Engel class I subgroups a to d demonstrates that patients who had a complete hippocampectomy were more likely to achieve an Engel class Ia outcome (P = .05 univariate and P = .04 multivariate, OR = 3.24, CI = 1.28,8.19). This significant relationship was also found among Engel class I patients with hippocampal sclerosis. With adjustment of length of follow-up and total neocortical resection, complete hippocampectomy was significantly associated with class Ia outcome (P = .03, OR = 3.82, CI = 1.17,12.54). Nonlesional (n = 49) cases demonstrated no difference in seizure outcome when comparing complete hippocampectomy with partial hippocampectomy
Potential differing factors in surgical approach between the partial and complete hippocampectomy groups were also evaluated. There was a difference in length of follow-up (3.7 years vs 7.4 years P < .001) with a longer follow-up in the partial resection group. The total extent of resection of lateral neocortex did not statistically correlate with Engel class I outcome (P = .93). A trend (Table 5) was noted for more extensive lateral neocortical resection in cases of class IV outcome (P = .18).
In our series, morbidity included the expected crescent superior quadrant visual field defects occasionally noted on examination. There were 2 cases of permanent dysphasia, 3 wound infections, and 2 deep vein thromboses. We did not observe any hemiparesis or hemianopia. There were no mortalities. Memory loss, cognitive dysfunction, and behavioral changes were not addressed, because the article is focused on the efficacy of surgery. These important factors are being evaluated and are in preparation to be reported.
For surgical candidates undergoing anterior temporal lobectomy, seizure freedom has been observed in 60% to 80% of cases.9,10,18–20 Hippocampal sclerosis is the most common pathological finding in adult temporal lobe epilepsy and has been independently identified as a positive predictor of excellent surgical outcome with amygdalohippocampectomy.8,18,21–25 Numerous factors have been predictive of outcome in this subgroup. Tassi et al22 demonstrated that hippocampal sclerosis, tumor resection, and lesion resection were correlated with good seizure outcomes and support that optimal outcomes are obtained when mesial and neocortical structures are removed. There is also a favorable rate of surgical success when there is detection of concordant abnormalities that indicate unilateral temporal lobe epilepsy in patients with nonlesional MRI findings.19,26 Our series also confirms that, like other methods, surgery for lesional pathology predicts Engel class I outcome.
The literature has shown that poor outcomes are seen with contralateral ictal discharges,27 diffuse or poorly localized epileptogenicity,28 generalized motor seizures,29 bilateral MRI abnormalities,30 and subdural electrodes.24,29 In our series the preresection factors that correlated with poor outcome were a history of secondary generalized seizures and the need for preoperative subdural electrode recording. Both may be surrogate markers for poor localization. Poor outcome with secondarily generalized seizures may be expected because this may mean the lateral neocortex is involved and not only mesial structures, making results variable.25,29
Many groups have published their surgical outcomes with the use of a variety of surgical approaches with similar results.17,31,32 Some literature has shown that transsylvian selective amygdalohippocampectomy appears to be favorable in reference to verbal and figural memory on the right side, but worse on the left side.17
In 2000, McKhann et al32 reported on a series of 140 patients who underwent a tailored temporal lobectomy in which the extent of hippocampal resection was determined intraoperatively by the extent of interictal epileptiform abnormalities on electrocorticography. In their study, the extent of hippocampal resection did not correlate with seizure control, and 67% of the patients were seizure free (Engel class I) with at least 18 months follow-up. However, persistent epileptiform discharges on the remaining hippocampus did correlate with persistent seizures and poorer outcome. In an article by Davies et al,33 anything less than class Ia was considered not seizure free. Wyler et al34 also reported on partial vs complete hippocampectomy, revealing significantly superior outcomes associated with total hippocampectomy when Engel class Ia (seizure free, aura free, no seizure at any time after surgery) was the outcome measure. However, this creates a discrepancy in reporting of the Engel class I (vs only class Ia) outcomes which confounds the literature. Our series addresses this discrepancy, and our findings are in agreement with both methods of reporting epilepsy outcomes. This may help to clarify the difference in the literature. The discrepancy lies in an evaluation of Engel class I subgroups. In our series, outcomes were reported as Engel class I vs other class outcomes, with a further analysis of Engel class Ia vs class Ib, c, and d. In our overall series and in the subgroup of patients with hippocampal sclerosis, with the tailored approach on multivariate analysis, complete hippocampal resection did not predict Engel class I outcome, but on analysis of Engel class subgroups a to d, patients who had a complete hippocampectomy were more likely to achieve an Engel class Ia outcome. This was not observed in cases with normal pathology. Thus, among patients with Engel class I outcome on multivariate analysis, complete hippocampectomy did correlate with a statistically significant lower incidence of persistent aura, drug withdrawal seizure, or early postoperative seizure. Although our data suggest that complete hippocampectomy may be the surgical strategy most likely to result in Engel class Ia outcome, this outcome must be balanced against the possible neuropsychological consequence of complete hippocampectomy in an individual patient as measured in careful preoperative testing.
Our data showed that total extent of resection of lateral neocortex did not statistically correlate with Engel class I outcome. This is similar to results seen in the literature with no difference between medial temporal and neocortical resections.35
Several limitations can be pointed out in our study. It is a retrospective analysis dealing with a heterogeneous group of patients. However, we have attempted to look at all possible confounding factors and adjust accordingly in our analysis. The retrospective analysis of multiple factors makes it difficult to be certain of the true significance of a positive finding, although the positive findings reported are in line with previous reports in the literature and thus lend further support to these previous reports. There was a mean follow-up of 5.4 years, but this varied among each group studied. This was true when looking at complete vs partial hippocampal resection in our class I subgroup analysis and when looking at the extent of lateral neocortex resection. However, the shorter length of follow-up in the complete hippocampectomy group is 3.7 years. This should be long enough for most patients to convert from Ia to Ib, c, or d if they are going to do so. The difference in long-term follow-up between the partial hippocampectomy group and the complete hippocampectomy group may be related to a larger number of nonlesional normal MRI cases earlier in the series. These normal MRI cases were more likely to have partial hippocampectomy. Later in the series, all patients with normal 1.5 T MRI findings underwent 3T MRI in which abnormalities were sometimes found and a complete hippocampectomy was more likely to be done. Longer-term follow-up on this important issue is planned. Our minimal follow-up period was 1 year and conforms to the literature presented for determining accurate seizure outcomes. Further analysis will need to be done to look at neuropsychological outcomes, and data specific to intraoperative electrocorticocography, as well. Last, mood effects like epilepsy-related depression have a large impact on quality of life and will need to be addressed. Follow-up analysis of all of these issues and their impact on clinical decision making is underway.
The results of this series of tailored temporal lobectomy demonstrate it as a safe and effective surgical method for temporal lobe epilepsy. Complications remained low in this series. Outcomes are similar to other techniques, with most patients (70%) achieving an excellent outcome of Engel class I regardless of a tailored approach. As in other methods, lesional pathology also predicts Engel class I outcome. Predictors of poor outcome were a history of secondary generalization and need for subdural electrode recording. The extent of lateral neocortical resection and the extent of hippocampal resection did not predict Engel class I outcome when the tailored approach was used. However, among Engel class I patients in the whole series and among the uniform group of patients with hippocampal sclerosis, complete hippocampal resection was predictive of Engel class Ia outcome. Therefore, it appears that partial hippocampectomy can achieve an Engel class I outcome, but radical hippocampectomy is more effective at leading to Engel class Ia outcome. The possible neuropsychological effects of radical vs partial hippocampectomy must be considered as well.
This article describes the seizure control outcomes in a large group of patients undergoing temporal lobectomy by a single surgeon with good long-term follow-up. It demonstrates once again that temporal lobe epilepsy surgery has a high success rate when performed on appropriately selected patients by an experienced team. The most interesting finding is the observation that patients with lesional epilepsy, including mesial temporal sclerosis, were more likely to achieve an Engel class Ia outcome (completely seizure free since surgery) if they underwent a complete rather than a partial hippocampectomy. This presents only half of the picture, however, because we are not provided with the results of pre- and postoperative neuropsychological testing. Therefore, we do not know whether there was a price to be paid for more aggressive hippocampal resection in terms of cognitive impairment, memory dysfunction, or behavioral changes. The authors indicate that their neuropsychological data are being collected for a future publication, and I look forward to this update.
All patients underwent intraoperative electrocorticography, and an estimated 5% had their resections extended as a result of the study. We are not given any direct or compelling evidence to suggest that this strategy resulted in better outcomes. The article does not help us decide whether electrocorticography should be used routinely to tailor temporal lobectomies, or whether patients would be served as well by a standard resection based on a careful preoperative assessment.
Andrew G. Shetter
The authors report the results of a retrospective study made on a rather large population of patients treated by "tailored" temporal lobe resection for drug-resistant focal epilepsy. In terms of seizure control, their results are comparable with those expected after surgery for temporal lobe epilepsy. The key point of the article is that in patients in Engel class I with hippocampal sclerosis, but not in nonlesional cases, complete hippocampectomy is associated with a significantly higher chance of Engel c lass Ia compared with partial hippocampectomy. Both persistent "auras" in cases with residual hippocampus at postoperative MRI, and significant cognitive impairment despite seizure freedom in cases with extensive hippocampal removal are common causes of postoperative disappointment in epilepsy surgery for the patient as well as for the surgeon. Actually, the choice of the most effective compromise in terms of both seizure control and cognitive preservation may not be simple in particular cases, especially those with an intact neuropsychological profile and normal MRI findings. Tailoring the surgical resection is the obvious solution in selected cases, and further clinical work, preferably in prospective multicentric studies, is required to develop a rationale to identify the different requirements and the resultant optimal choice.
- epilepsy, generalized
- objective (goal)
- hippocampus (brain)
- preoperative care
- subdural space
- wound infection
- temporal lobe
- visual field defects
- predictor variable
- hippocampal sclerosis
- refractory epilepsy
- secondarily generalized seizures