OUP user menu

Surgical outcome in patients with epilepsy and dual pathology

L. M. Li, F. Cendes, F. Andermann, C. Watson, D. R. Fish, M. J. Cook, F. Dubeau, J. S. Duncan, S. D. Shorvon, S. F. Berkovic, S. Free, A. Olivier, W. Harkness, D. L. Arnold
DOI: http://dx.doi.org/10.1093/brain/122.5.799 799-805 First published online: 1 May 1999

Summary

High-resolution MRI can detect dual pathology (an extrahippocampal lesion plus hippocampal atrophy) in about 5–20% of patients with refractory partial epilepsy referred for surgical evaluation. We report the results of 41 surgical interventions in 38 adults (mean age 31 years, range 14–63 years) with dual pathology. Three patients had two operations. The mean postoperative follow-up was 37 months (range 12–180 months). The extrahippocampal lesions were cortical dysgenesis in 15, tumour in 10, contusion/infarct in eight and vascular malformation in five patients. The surgical approach aimed to remove what was considered to be the most epileptogenic lesion, and the 41 operations were classified into lesionectomy (removal of an extrahippocampal lesion); mesial temporal resection (removal of an atrophic hippocampus); and lesionectomy plus mesial temporal resection (removal of both the lesion and the atrophic hippocampus). Lesionectomy plus mesial temporal resection resulted in complete freedom from seizures in 11/15 (73%) patients, while only 2/10 (20%) patients who had mesial temporal resection alone and 2/16 (12.5%) who had a lesionectomy alone were seizure-free (P < 0.001). When classes I and II were considered together results improved to 86, 30 and 31%, respectively. Our findings indicate that in patients with dual pathology removal of both the lesion and the atrophic hippocampus is the best surgical approach and should be considered whenever possible.

  • hippocampal sclerosis
  • dual pathology
  • epilepsy
  • surgery
  • outcome

Introduction

In patients with partial epilepsy and an underlying epileptogenic lesion it is likely that seizures originate from that lesion or its vicinity (Gloor, 1987). Recognition of the underlying lesion and definition of the extent of the epileptogenic area are key points in the presurgical evaluation of patients with refractory partial epilepsy (Gloor, 1987; Awad et al., 1991; Cascino et al., 1993a). Surgical removal of these epileptogenic areas can be curative or provide significant reduction in seizure frequency in the majority of individuals.

The most common pathological finding in surgical specimens of patients with refractory partial seizures is hippocampal sclerosis, followed by tumour, cortical dysgenesis and vascular malformation (Babb and Brown, 1987; Engel, 1993).

Hippocampal sclerosis can now be diagnosed in vivo by its morphometric and signal intensity changes on MRI by two-dimensional visual inspection (Jackson et al., 1994; Bronen et al., 1995), by quantitative measurements, such as volumetry (Jack et al., 1990; Cook et al., 1992; Cendes et al., 1993; Jack et al., 1995a; Watson et al., 1996a) or by T2 relaxometry (Van Paesschen et al., 1995). Strong correlation has been found between hippocampal volume and pathological findings in different centres (Cascino et al., 1991; Lencz et al., 1992; Lee et al., 1995; Watson et al., 1996b; Van Paesschen et al., 1997). Surgical removal of a unilateral atrophic hippocampus can render >80% of patients seizure-free (Arruda et al., 1996) while surgical outcome is less favourable in patients with bilateral atrophy (Jack et al., 1995b; Arruda et al., 1996) and in those without atrophy (Berkovic et al., 1995; Jack et al., 1995b; Arruda et al., 1996).

The surgical outcome in lesional epilepsy is variable and good results range from 39 to 83% (Cascino et al., 1993a). In our previous study (Li et al., 1997a), we found that incomplete removal of the lesion appeared to be the main reason for an unfavourable surgical result. In patients with complete removal of the macroscopic lesion, 85% became seizure-free or had an important reduction in seizure frequency (Engel's class I–II).

In some patients, however, MRI can show the association of hippocampal atrophy with an extrahippocampal lesion, defined as dual pathology. The frequency of this association ranges from 5 to 30% in patients with refractory partial epilepsy (Levesque et al., 1991; Cascino et al., 1993b; Raymond et al., 1994; Cendes et al., 1995; Li et al., 1995). This wide range in the reported frequency of dual pathology reflects the different populations being studied and the methods of analysis. In quantitative MRI-based studies, the frequency of dual pathology was more homogeneous and found to be ~15% (Cascino et al., 1993b; Raymond et al., 1994; Cendes et al., 1995). The most common types of extrahippocampal lesions found in dual pathology are developmental abnormalities, such as cortical dysgenesis, followed by gliotic lesions acquired in early childhood (Levesque et al., 1991; Raymond et al., 1994; Cendes et al., 1995).

There have been very few reports suggesting the optimal surgical strategy in individuals with dual pathology. Cascino et al. (1993) reported three patients who became seizure-free only after a second operation in which the atrophic hippocampus was removed. In our previous study of surgical outcome in patients with single and dual pathology (Li et al., 1997a) we concluded that, in patients with dual pathology, both the extrahippocampal lesion and the atrophic hippocampus were likely to be involved in seizure generation. The best result was obtained when both the extrahippocampal lesion and the atrophic hippocampus were removed. However, because of the small number of patients with dual pathology (Li et al., 1997a), we could not conclude whether the type of lesion or the degree of hippocampal atrophy is the most important factor when the surgical strategy is being decided. In the multicentre study reported here we extended our analysis to a much larger group of patients with dual pathology who underwent surgical treatment for refractory epilepsy.

Patients and methods

We used as the criterion for inclusion concomitant hippocampal sclerosis and an extrahippocampal lesion diagnosed either by pathology or by MRI when the lesion was not resected. The patients were from five centres: the Montreal Neurological Institute and Hospital, Canada; the National Hospital for Neurology and Neurosurgery, London; Wayne State University, Detroit, Michigan, USA; the Royal Melbourne Hospital, Australia; and the Austin and Repatriation Medical Center, Victoria, Australia. All had a postoperative follow-up of at least 1 year.

Thirty-eight patients (15 women) with a mean age of 31 years (range 14–63 years) were studied (Table 1). The mean age of seizure onset was 13 years (range 1–31 years). Mean postoperative follow-up was 37 months (range 12–180 months; seven patients had a follow-up of <2 years).

Hippocampal sclerosis was defined either by preoperative MRI volumetric measurement of the hippocampal formation (Cook et al., 1992; Cendes et al., 1993; Watson et al., 1997) or by pathology (Babb and Brown, 1987; Meencke and Veith, 1991).

MRI volumetric measurements of the hippocampal formation were performed using 1–3 mm thick coronal slices following a specific protocol (Cook et al., 1992; Cendes et al., 1993; Watson et al., 1997). Forty-four age-matched normal control individuals were scanned using the same protocol. Volumetric studies from the centres were validated by cross-measurement (Cendes et al., 1992). The volume of the hippocampal formation in the control group had a normal distribution curve. We compared the volumes of hippocampal formation of the patients with those of the control group, and analysed the amount of asymmetry between sides, expressed as the ratio of the smaller to the larger hippocampal volume. Values of <0.94 (<2 SD) from the mean of the control group were considered abnormal. Values of <0.75 were considered as indicating severe hippocampal sclerosis (Watson et al., 1996).

Histopathological diagnosis of hippocampal sclerosis was based on qualitative and descriptive analysis. Sommer's sectors (CA1 to CA4) showed a pattern in keeping with hippocampal sclerosis, and because no quantitative cell counting was performed a broader grading system was chosen: (i) mild to moderate hippocampal sclerosis (mild or moderate astrogliosis and neuronal loss in the hippocampal specimen); and (ii) severe hippocampal sclerosis (severe neuronal loss in the hippocampus and pronounced proliferation of astrocytes involving more than one medial temporal structure). The term hippocampal sclerosis in the present study reflects a spectrum of neuronal cell loss and astrogliosis in the mesial temporal structures. This includes not only the `classical hippocampal sclerosis' (Mathern et al., 1997) but also mild degrees of hippocampal neuronal loss and astrogliosis of mesial temporal structures.

Extrahippocampal lesion classification was based on the pathology described in the surgical specimen. When the lesion was not included in the resection it was classified according to the MRI findings using spin echo (T1- and T2-weighted), inversion recovery and gadolinium–DTPA enhanced images. The lesions were all visible on MRI.

Prolonged scalp EEG recordings were obtained using the International 10–20 system and sphenoidal electrodes. Interictal EEGs showed epileptiform activity over the temporal regions in 34/38 patients, predominating over the side of hippocampal sclerosis. Independent interictal epileptiform activity was also seen over the underlying lesion in 30/38 patients. Ictal EEG was recorded in 36/38 patients. Six patients with coexistent periventricular nodular heterotopia had chronic intracranial recording for further seizure localization/lateralization; however, the nodules were not implanted.

The surgical decision was based on all available clinical information, prolonged video-EEG monitoring and results of other complementary tests. The most likely origin of the seizures, as defined by neurophysiological studies, was the target of the resection. Three patients had two operations: 41 surgical procedures were analysed with respect to outcome (Table 2). The 41 operations were classified into three categories: lesionectomy (removal of the extrahippocampal lesion); mesial temporal resection (removal of the atrophic hippocampus); and lesionectomy plus mesial temporal resection (removal of both the lesion and the atrophic hippocampus).

Surgical outcome was based on Engel's modified classification (Engel et al., 1993): class I, seizure-free or residual auras; class II, rare seizures (fewer than three complex partial seizures per year); class III, significant improvement with 90% reduction in seizure frequency; class IV, no worthwhile improvement.

We assessed the relative homogeneity in the distribution of severity of hippocampal sclerosis and of the types of lesion among different types of surgical procedures. Using either histology or MRI volumetric studies to define severe hippocampal sclerosis, 11/15 (73%) of patients who had lesionectomy plus mesial temporal resection, 5/10 (50%) of patients who had mesial temporal resection and 7/16 (44%) of patients who had lesionectomy had severe hippocampal sclerosis. There was no significant association between the severity of hippocampal sclerosis and the type of operation performed [χ2(2) = 2.95, P = 0.29]. Dividing the type of lesion into those with `better' (tumours and vascular malformations) and those with `poorer' prognosis (cortical dysgenesis, infarct and contusions) in regard to surgical outcome, 8/15 (53%) of patients who underwent lesionectomy plus mesial temporal resection, 8/10 (80%) of those who underwent mesial temporal resection and 9/16 (56%) of patients who underwent lesionectomy had a `poorer prognosis' type of lesion. This was also not significant [χ2(2) = 2.04, P = 0.36].

We used multivariate ANOVA (analysis of variance) and simple main effect and post hoc pairwise comparison in order to assess whether the type of surgery (lesionectomy versus mesial temporal resection versus lesionectomy plus mesial temporal resection), the type of lesion (better versus worse prognosis) and the degree of hippocampal sclerosis (mild to moderate versus severe) might have an independent or an interactive effect on the surgical outcome. Fisher's exact test was used to assess the significance of the association between ictal EEG localization and surgical outcome.

Results

Freedom from seizures was achieved in 11/15 (73%) of patients who had a lesionectomy plus mesial temporal resection compared with only 2/10 (20%) of those who had a mesial temporal resection alone and 2/16 (12.5%) of those who had a lesionectomy alone. When classes I and II were considered together the percentages rose to 86, 30 and 31%, respectively.

Multivariate ANOVA showed that the surgical outcome was significantly affected by the type of surgical procedure [F(1,40) = 9.02, P = 0.0009] and that there was a significant interaction between the type of the surgical procedure and the degree of hippocampal sclerosis [F(2,40) = 5.62, P = 0.008]. Surgical outcome was not significantly affected by the type of lesion [F(1, 40) = 0.29, P = 0.592] or by the degree of hippocampal sclerosis [F(1,40) = 0.19, P = 0.666]. There was no significant interaction between the type of lesion and either the type of surgical procedure [F(2,40) = 0.41, P = 0.666] or the degree of hippocampal sclerosis [F(1,40) = 1.07, P = 0.309]. There was also no significant interaction between the type of lesion, the type of surgical procedure and the degree of hippocampal sclerosis [F(2,40) = 0.95, P = 0.4].

Simple main effect analysis, seeking the significant factor in the interaction between the type of surgery and the degree of hippocampal sclerosis on the surgical outcome, showed that the type of surgery was a significant factor both for patients with severe hippocampal sclerosis [F(2,29) = 7.35, P = 0.002] and for those with mild to moderate sclerosis [F(2,29) = 7.11, P = 0.003].

Post hoc comparison, using Tukey's honestly significant difference test, showed that there were significant differences between lesionectomy plus mesial temporal resection and both lesionectomy alone [q(29) = 7.67, P < 0.01] and mesial temporal resection alone [q(29) = 5.02, P < 0.01], but there was no significant difference between lesionectomy alone and mesial temporal resection alone [q(29) = 1.75, P > 0.05].

Simple main effect analysis of the degree of hippocampal sclerosis with respect to the outcome from different types of surgery showed that the degree of hippocampal sclerosis had a significant effect in patients who had mesial temporal resection alone [F(1,29) = 8.20. P = 0.007] but not in those who had a lesionectomy alone [F(1,29) = 3.11, P = 0.088] or in those who had lesionectomy and mesial temporal resection combined [F(1,29) = 0.12, P = 0.732. Three of five patients with severe hippocampal sclerosis treated by mesial temporal resection alone achieved a class I–II outcome compared with 0/5 of those with mild to moderate hippocampal sclerosis.

Two out of seven patients (28%) with localized ictal EEG and 3/9 (33%) with non-localized ictal EEG who underwent lesionectomy alone were in class I–II. This was not statistically significant (Fisher's exact test, P = 1, two-tailed)

One of the three patients (33%) with localized ictal EEG and 2/7 (28%) with non-localized ictal EEG who underwent surgical removal of the sclerotic hippocampus alone were in class I–II. This did not achieve statistical significance (Fisher's exact test, P = 1, two-tailed).

Discussion

Ideally one would like to carry out a randomized prospective study of the three types of surgical procedure. Nevertheless, this study is of interest because it evaluates the result of what would be the `optimal' individual procedure in the light of current knowledge and technology for seizure localization from five centres (in Canada, USA, UK and Australia).

Cascino et al. (1993b) reported three patients with dual pathology who had an unsuccessful outcome after lesionectomy (two vascular malformations and one ganglioglioma); they achieved freedom from seizures following a second operation in which the atrophic hippocampus was resected. In our series, the best results were seen in those patients who had both the lesion and the atrophic hippocampus removed. Seventy-three per cent became completely seizure-free compared with 20% of those who had resection of the atrophic hippocampus alone and 12.5% of those who had only a lesionectomy. The best outcome in this series was 86%, when both class I and class II were included (Table 2). This is comparable to the results obtained in mesial temporal sclerosis (90% in class I–II) (Arruda et al., 1996) or in lesional epilepsy syndrome (85% in class I–II) (Li et al., 1997a).

In the present series, the results of prolonged scalp EEG recording, whether localizing/lateralizing or not, showed no significant association with outcome. Our results indicate that in the presence of dual pathology both the lesion or its surround and the atrophic hippocampus are potentially epileptogenic. Invasive recording might provide further clarification about which component is the most epileptogenic. It is still unknown whether the risk : benefit ratio of such an invasive approach could significantly refine the surgical strategy and ultimately improve the outcome.

In patients who had resection of medial temporal structures alone, we found better surgical outcome in those with severe hippocampal sclerosis compared with those with mild to moderate hippocampal sclerosis (3/5 versus 0/5 in class I–II). Six of 10 patients in this group who had resection of the atrophic hippocampus alone had coexistent periventricular nodular heterotopia, and all of them were in class III–IV. The exact role of grey matter nodules in the epileptogenicity in these cases is not entirely clear. We (Li et al., 1997b) reported a series of patients with refractory temporal lobe epilepsy and periventricular nodular heterotopia who had poor surgical results after temporal lobe removal. Possible explanations for the poor surgical results seen in these patients included: (i) periventricular nodular heterotopia as the epileptogenic source; (ii) dual pathology; (iii) periventricular nodular heterotopia being part of a more widespread epileptogenic source. Electrophysiological recording (Dubeau et al., 1997; Kothare et al., 1998) from the nodules provided support for the first two hypotheses.

In the group of patients who had removal of the lesion alone, the severity of hippocampal atrophy was not associated with either good or bad surgical outcome (Fig. 1). This suggests that the presence of hippocampal atrophy, independently of its severity, may have a negative effect on the surgical result. This is of clinical relevance because two-dimensional visual inspection may not reliably diagnose subtle hippocampal atrophy. Therefore, especially in patients with early developmental lesions referred for presurgical evaluation, a careful examination of the mesial temporal structures with quantitative measurement should be considered in view of the known association of such lesions with hippocampal sclerosis (Levesque et al., 1991; Raymond et al., 1994; Cendes et al., 1995).

In conclusion, assessment of possible dual pathology in patients with lesional epilepsy is important because their surgical management and outcome are distinct. For patients with dual pathology, removal of both the lesion and the atrophic hippocampus should be considered whenever possible, preferably in one step when the coexistent lesion is in the temporal lobe. For patients with a lesion outside the temporal lobe, the decision of whether the lesion or the atrophic hippocampus should be removed first should be based on the location, extent and nature of the lesion. Factors such as malignanleftcy and the potential risk of bleeding must be considered. A two-step intervention could be planned and explained to the patient, taking into consideration the individual requirements and goals of surgical treatment.

View this table:
Table 1

Summary of clinical data in 38 patients with dual pathology

Age/age of seizure onset (years)/sexIctal EEGType of extrahippocampal leisonRatio S/LDegree of hippocampal sclerosisType of resectionFollow-up(months)Outcome*
M = male; F = female; R = right; L= left; Bi = bilateral; F = frontal; P = parietal; C = central; T = temporal; O = occipital;S/L = small/larger hippocampus; MT = mesial temporal resection; LS = lesionectomy. *Engel's modified classification (see text for details).
38/2/FR CTR TPO cortical dysgenesisSevereMT + part LS40I
30/17/ML TL T cortical dysgenesis0.88InconclusiveMT + part LS50I
19/1/MNo changeL TPO contusion0.72InconclusiveMT + part LS25I
16/2/MBi FR TO infarct0.41SevereMT + part LS12III
59/22/FNot recordedR T gangliogliomaSevereMT + LS48I
35/26/MR TR T gangliogliomaSevereMT + LS60I
34/18/ML TL T gangliogliomaSevereMT + LS84I
31/16/ML TL T vascular malformationModerateMT + LS48I
14/1/ML TL T cortical dysgenesisSevereMT + LS36I
29/18/FNot recordedL T vascular malformation0.76ModerateMT + LS14II
63/17/ML TL T vascular malformationMildMT + LS48I
43/13/FBi T L > RL T contusion0.75Not availableMT + LS19I
28/12/FR TBi periventricular nodular heterotopia0.91ModerateMT31III
23/17/MBiR periventricular nodular heterotopiaSevereMT53III
18/2/ML TL T dysembryoplastic neuroepithelial tumourSevereMT24I
38/25/MBi R > LL insula vascular malformationModerateMT24IV
37/13/ML TBi periventricular nodular heterotopiaMildMT24III
36/10/FL CT/POL P cortical dysgenesis0.73InconclusiveMT18II
33/28/FR hemisphereBi periventricular nodular heterotopiaModerateMT60IV
27/16/MBi FTR periventricular nodular heterotopiaSevereMT56IV
27/7/MBiBi periventricular nodular heterotopiaMild–moderateMT42III
34/15/FL FL F cortical dysgenesis0.64SevereMT17I
15/9/MBiR FTP cortical dysgenesis0.79LS (part)16IV
39/7/FL FCTL F contusion0.89LS57III
36/26/FNo changeL F astrocytoma0.71LS34III
50/31/MR TR T dysembryoplastic neuroepithelial tumour0.87LS12I
30/19/ML TFR F orbital and L T polar epidermoid tumour0.86LS26II
20/15/ML TL T cortical dysgenesis0.89LS24IV
16/1/FL hemisphereBi T cortical dysgenesis0.78LS24II
41/2/FL TR F meningioma0.66LS180IV
37/13/ML TL T dysembryoplastic neuroepithelial tumour0.45LS31III
32/1/ML TL T contusion0.84LS12IV
28/17/FR TR TO cortical dysgenesis0.88LS33II
23/8/FBi FL FT contusion0.87LS17I
18/14/ML TL T ganglioglioma0.86LS35IV
30/10/ML TL T vascular malformationSevere1st LS, 2nd LS + MT40III
36/19/MR FTR F contusionSevere1st LS, 2nd LS + MT36I
18/1/FR TR P contusion0.75Inconclusive1st LS, 2nd LS + MT18I
View this table:
Table 2

Different surgical approaches and outcome in patients with dual pathology

LesionectomyResection of atrophic hippocampusLesionectomy plus resection of atrophic hippocampus
*One patient had rare auras; †six of seven patients had associated periventricular nodular heterotopia; ‡multivariate analysis of variance: F(1,40) = 9.02, P = 0.0009.
Class I 2212*
Class II 31 1
Class III–IV117† 2
Seizure-free/total 2/16 (12.5%)2/10 (20%)11/15 (73%)‡
Class I–II/total 5/16 (31%)3/10 (30%)13/15 (86%)‡
Fig. 1

37-year-old man with dual pathology and seizures since the age of 13 years. (A) Axial T1-weighted MRI with gadolinium showed a left temporal lesion. He underwent lesionectomy and histopathology showed dysembryoplastic neuroepithelial tumour. (B) Postoperative coronal T1-weighted MRI showed left hippocampal atrophy (ratio smaller/larger = 0.70). The hippocampus was left in place. The patients was in class III with 31 months follow-up.

Acknowledgments

Dr Li is supported by the Savoy Foundation for Research in Epilepsy of Canada.

References

View Abstract