SEARCH

SEARCH BY CITATION

Keywords:

  • Atypical hippocampal sclerosis;
  • Granule cell dispersion;
  • Surgical outcome

Summary

  1. Top of page
  2. Summary
  3. Methods and Materials
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure
  8. References

Purpose:  Around one-third of patients undergoing temporal lobe surgery for the treatment of intractable temporal lobe epilepsy with hippocampal sclerosis (HS) fail to become seizure-free. Identifying reliable predictors of poor surgical outcome would be helpful in management. Atypical patterns of HS may be associated with poorer outcomes. Our aim was to identify atypical HS cases from a large surgical series and to correlate pathology with clinical and outcome data.

Methods:  Quantitative neuropathologic evaluation on 165 hippocampal surgical specimens and 21 control hippocampi was carried out on NeuN-stained sections. Neuronal densities (NDs) were measured in CA4, CA3, CA2, and CA1 subfields. The severity of granule cell dispersion (GCD) was assessed.

Results:  Comparison with control ND values identified the following patterns based on the severity and distribution of neuronal loss: classical HS (CHS; n = 60) and total HS (THS; n = 39). Atypical patterns were present in 30% of cases, including end-folium sclerosis (EFS; n = 5), CA1 predominant pattern (CA1p; n = 9), and indeterminate HS (IHS, n = 35). No HS was noted in 17 cases. Poorest outcomes were noted for no-HS, and CA1p groups with 33–44% International League Against Epilepsy (ILAE) class I at up to 2 years follow-up compared to 69% for CHS (p < 0.05). GCD associated with HS type (p < 0.01), but not with outcome.

Conclusions:  These findings support the identification and delineation of atypical patterns of HS using quantitative methods. Atypical patterns may represent distinct clinicopathologic subtypes and may have predictive value following epilepsy surgery.

Surgery is regarded as the treatment of choice in the management of patients with intractable mesial temporal lobe epilepsy (MTLE) due to hippocampal sclerosis (HS) (Wiebe et al., 2001). The literature provides relatively consistent data that up to approximately two-thirds of patients will become seizure-free after surgery (Engel, 1996; Sperling et al., 1996; Foldvary et al., 2000; Mcintosh et al., 2001; Spencer & Huh, 2008). There is some evidence that with longer follow-up periods of 5 years, the number of patients remaining seizure free declines (Aull-Watschinger et al., 2008). What is perplexing is why patients with similar seizure type, severity, and similar pathology respond differently to surgery, with a subgroup of these carefully selected patients failing to become seizure-free. Possible causes include incomplete resection and the presence of a more extensive epileptogenic network involving the temporal and extratemporal regions beyond the resected lesion, including the contralateral hippocampus (Spencer & Huh, 2008; Bertram, 2009).

Predictors of poor postoperative outcome are desirable and have been sought from analysis of clinical parameters, for example, early recurrence of seizures in the postoperative period (Aull-Watschinger et al., 2008; Spencer & Huh, 2008; Tezer et al., 2008). There is also the possibility that patterns of hippocampal damage observed in the resected tissue can provide predictive information. For example “classical” patterns of neuronal loss involving primarily CA1 and CA4 subfields may be predictive of better outcome than “atypical” HS with focal or restricted neuronal loss (de Lanerolle et al., 2003; Blumcke et al., 2007). Indeed HS is unlikely to be a single entity; for example, different patterns of HS may be confirmed at postmortem examination in epilepsy patients with a variety of syndromes (Thom et al., 2009). Therefore, subtle differences in the patterns of hippocampal neuronal loss in patients with MTLE may also reflect different pathoetiologies and clinical subtypes. The aim of this study was to replicate previous study findings through a retrospective analysis of patterns of HS in our series of operated patients.

Methods and Materials

  1. Top of page
  2. Summary
  3. Methods and Materials
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure
  8. References

Case and tissue selection

Cases were obtained from the archives of the neuropathology department of the National Hospital for Neurology and Neurosurgery. This study has been approved by the joint local ethics committee of the Institute of Neurology and National Hospital for Neurology and Neurosurgery. Individual patient consent was obtained in accordance. All patients had undergone anterior temporal lobectomy and hippocampectomy during adulthood for the treatment of refractory temporal lobe epilepsy. Cases were included from the operative period 1994–2006 and were included on the basis of availability of archival hippocampal material, with at least three hippocampal subfields represented in the tissue (including CA4 and CA1 sectors) and where sufficient clinical follow-up data were available. Cases with second distinct lesional pathology (e.g., cavernoma, tumour, focal cortical dysplasia) were not included. One hundred sixty-five hippocampal specimens met the study requirements (age range 17–60 years; mean 34 years).

A single block from each case was selected through the mid body of the hippocampus, a 7-μm section cut, and routine NeuN immunohistochemistry applied (1:2,000; Chemi-Con, Temecula, CA, U.S.A.; microwave pretreatment, 1 h incubation). Postmortem control hippocampi from 14 patients without a history of epilepsy or other neurologic illness were included (age range 33–63 years at death; mean age 47 years). A single block from the mid body of the left and right sides was used where available, resulting in 21 hippocampal specimens. NeuN staining was carried out on the postmortem tissue using a modified immunostaining method (1:500, formic acid pretreatment, overnight incubation) to ensure optimal detection of neuronal nuclei (Gill et al., 2005).

Quantitative methods

The quantitative method employed was based on a recently published study (Blumcke et al., 2007). Neurons were analyzed using Image Analyser (Histometrix, Kinetic Imaging, Nottingham, United Kingdom) utilizing a Zeiss Axioskop microscope (Carl Zeiss, Oberkochen, Germany). Regions of interest (ROIs) in each of subfields CA4, the stratum pyramidale of CA3, CA2, and CA1, were outlined using a ×2.5 objective lens. For the CA4 region, the entire area within the blades of the dentate gyrus was outlined in each case. This region included cells of the polymorphic layer in addition to hilar pyramidal neurones. Within these ROIs, 10 random fields at objective ×40 were examined in each of CA2, CA3, and CA4, and 20 fields in CA1. All NeuN-positive neurons, regardless of size or morphology, were counted in each field, and the neuronal density for each subfield was calculated. Similar analysis was carried out on both epilepsy tissue and controls. The total time for quantitative analysis of one section varied between 30 and 60 min per case.

Z Scores were then calculated for each subfield in each case. The z score reflects the deviation of a subfield density from the predetermined population mean, in this case the mean autopsy subfield values, and is calculated by subtracting the mean control value and dividing by the standard deviation. In this study only negative z scores of magnitude 2 or greater were considered relevant; positive scores were interpreted as overlapping with normal autopsy control values. From the z scores the following patterns of HS were determined: normal or no-HS (z score of magnitude <2 in each subfield), classical HS (CHS) (z score <2 in CA2 but >2 in all remaining subfields), total HS (THS) (z scores of >2 in all subfields including CA2), CA1 predominant HS (CA1p) (z score >2 in CA1 but <2 in CA4), end-folium sclerosis (EFS) (z score of >2 in CA4 only), and indeterminate HS (significant z scores of 2 or more in at least one subfield not falling into one of the above patterns) (Fig. 1). Cases with no CA2 available for analysis but significant neuronal loss in remaining subfields were classified as CHS. In addition, with blinding to the quantitative evaluation, a qualitative assessment of the pattern of HS was made (CHS, THS, EFS, or CA1p pattern) based on the NeuN as well as conventionally stained sections [hematoxylin and eosin (H&E) and glial fibrillary acid protein (GFAP)] by two observers independently (KE/MT).

image

Figure 1.   Patterns of hippocampal sclerosis (HS) and granule cell dispersion (GCD). (A) Method of estimating the degree of granule cell dispersion in HS cases as detailed in methods section. Using image analysis, “best fit” lines were drawn (shown in yellow) for the most distal and basal granule cells, and the mean distance between these two lines measured (shown in red). This was repeated in eight different fields and the average values per case recorded (GCD mean). (B) CHS (classical hippocampal sclerosis): Significant neuronal loss (z score > 2) was detected in CA1 and CA4 with preservation of neurons in CA2 and subiculum (SC). (C) CA1p (CA1 predominant neuronal loss). Significant levels of neuronal loss were detected only in CA1 subfield with better preservation of neurons in CA4, CA3, and CA2. The granule cell layer is noted to be compact in this case (SC = subiculum). (D) EFS (end folium sclerosis). Neuronal loss in CA4 (arrowed) only, with conservation of neurons in CA1. Although there are some artifacts in the CA3 region, there is neuronal preservation in CA3 and CA2. (E) Indet (indeterminate HS). Significant neuronal loss in CA3 and CA1 in this case, but relative preservation in CA4. Because this does not fall into one of the predefined patterns of HS, this was classified as an indeterminate HS. (SC = subiculum). (F) THS (Total hippocampal sclerosis). Significant neuronal loss in all subfields, including CA2. Note preservation of neurons in subiculum (SC). All are NeuN-stained sections; Bars in (A) 50 μm, (B–F) 1,000 μm.

Download figure to PowerPoint

In the dentate gyrus, neuronal numbers were not counted because the presence of granule cell dispersion (GCD), present in more than 40% of HS cases (Wieser, 2004), can invalidate numerical density measures as an estimation of neuronal loss even with stereologic methods (Lurton et al., 1997; Van Paesschen et al., 1997; Thom et al., 2002). Measurements of the width of the granule cell layer (GCL), as an estimate of (GCD), were, therefore, carried out by a single observer (MT). Using Image Pro plus software (Media Cybernetics, Marlow, United Kingdom) on the same NeuN sections, images were captured using a ×20 objective lens along the granule cell layer. Images were taken from the regions with the impression of maximal GCD to include both the inner and outer blades of the GCL. In cases with granule cell depletion the images were taken from any remaining regions of the GCL. In each image the 10 most distal granule cells in the outer molecular layer were tagged, and “best fit” straight line between these points was drawn. A similar process was carried out with innermost cells at the hilar border of the granule cell layer. The average distance between the two drawn lines was measured (Fig. 1A). This measurement was repeated in eight regions along the GCL per slide, and both the mean (GCD mean) and maximum GCD (GCD Max) per case was recorded. This method, including image capture, was repeated in five random cases with good reproducibility of data.

Patient notes were reviewed for the preoperative epilepsy history, in particular age of onset and duration of epilepsy to surgery and age at and type of any initial precipitating injury or event (IPI). Postoperative follow-up information was obtained at 2 years for all but 15 patients for whom follow-up was available for 1 year only. The ILAE system was used for evaluation of seizure control postsurgery (Wieser et al., 2001). Statistical analysis was carried out using SPSS version 16 for windows; p values of <0.05 were taken as significant between groups.

Results

  1. Top of page
  2. Summary
  3. Methods and Materials
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure
  8. References

Quantitative analysis revealed the following patterns of HS in this series of 165 cases: CHS (36%), THS (24%), EFS (3%), CA1p (6%), indeterminate (21%), and no-HS (10%) (Fig. 1, Table 1). This was based on quantitative evaluation of four subfields in the majority of sections including CA2 subfield, which was available for analysis in 84% and CA3 region in 79% of sections. Within the indeterminate HS group, significant z scores (>2) were observed in 35 cases, which in 18 cases involved CA4 and CA3 alone.

Table 1.   Clinical and pathology measurement data in HS subgroups
Pattern HS n = number of cases (% of total cases)Patient age at surgery/PM Mean (SD) yearsAgreement with qualitative assessment (%)Neuronal densities × 10−4/μm2 CA1 (SD) CA2 (SD) CA3 (SD) CA4 (SD)Granule cell dispersion Mean (SD) Maximum (SD) (μ)Mean age at IPI (SD) yearsMean age of habitual seizure onset (SD) yearsMean duration of seizures (SD) years
  1. HS, Hippocampal sclerosis; PM, postmortem; IPI, initial precipitating injury; CA1p type, CA1P predominant sclerosis; EFS, end folium sclerosis; CHS, classical hippocampal sclerosis; THS, total hippocampal sclerosis. Statistical analysis carried out between HS groups (or between HS groups and controls)* with ANOVA. N/A, not applicable (an indeterminate pattern was only defined following quantitative evaluation).

CA1p type n = 9 (6%)32.5 (7.2)1000.56 (0.18) 2.66 (0.6)  2.20 (0.37) 1.81 (0.37)124 (55)  182 (69) 1.8 (2.3)  5.9 (5.8)26.5 (7.7) 
EFS n = 5 (3%)33.4 (5.2)60 2.6 (1.34)  2.9 (0.8)   1.4 (1.04) 0.37 (0.2) 183 (68)  269 (110)2.4 (1.24)18.2 (5.0)15.2 (5.6) 
CHS n = 60 (36%)33.6 (8.1)770.48 (0.21) 1.77 (1.04) 0.84 (0.46) 0.35 (0.23)213 (79)  291 (111)1.7 (1.4)  9.3 (8.1)24.2 (8.9) 
THS n = 39 (24%)31.9 (8.6)530.40 (0.17) 1.16 (0.47) 0.62 (0.34) 0.39 (0.23)194 (87)  260 (118)1.9 (2.5)  6.5 (6.4)25.5 (10.0)
Indeterminate n = 35 (21%)34.7 (8.7)N/A1.29 (1.22) 1.65 (0.8)  1.02 (0.72)  0.8 (0.54)173 (79)  228 (98) 3.7 (6.0) 10.8 (7.1)24.0 (9.8) 
Normal (No HS) n = 17 (10%)35.4 (8.6)352.82 (1.60) 3.12 (1.55) 2.75 (0.90) 2.07 (0.92)132 (41)  190 (67) 1.2 (0.8) 12.4 (8.3)23.1 (6.5) 
Controls n = 2149 (10.1) 3.08 (1.25) 4.22 (0.99) 3.45 (0.82) 1.97 (0.61) 79 (16)  127 (29) 
Significance between HS groups (HS and controls)*p < 0.05*p < 0.0001p < 0.001 p < 0.01p = 0.181p < 0.01p = 0.26

Poorer seizure-free outcomes were noted for the atypical patterns CA1p, and indeterminate HS and significant differences in outcome were noted between the HS groups (p = 0.043) (Fig. 2). Poorer outcomes were noted for the no-HS group, with only 44% seizure-free (ILAE class 1) (Wieser et al., 2001) compared to 100% and 71% in the EFS and THS groups.

image

Figure 2.   Outcome at 2 years follow-up using ILAE criteria (grouped as grade 1, 2, and 3, or greater) for different patterns of sclerosis. There was a significant difference in outcome between the groups [analysis of variance (ANOVA), p = 0.043) corrected for multiple comparisons. Indet., indeterminate; no HS, no hippocampal sclerosis; CA1p, CA1 predominant sclerosis; EFS, end folium sclerosis; CHS, classical hippocampal sclerosis; THS, total hippocampal sclerosis.

Download figure to PowerPoint

There were differences in mean duration of epilepsy between the HS groups (Table 1), with shortest duration to surgery for EFS and no-HS cases but this was not significant. Variation in the age of onset of habitual seizures was also noted, with mean age of onset in the CA1p group of 5.9 years spanning to 18.2 years in EFS, which was statistically significant (p < 0.01) (Table 1).

Data regarding an IPI or not preceding the onset of habitual epilepsy were recorded in 162 patients; in 38 there was no history of an IPI (Table 2). The age at IPI, regardless of type, was not significantly different between HS groups (Table 1). The most common IPI was a febrile seizure in 51% of patients overall, but reported in only 35% of patients with no-HS. The lack of a history of any IPI was noted in more than half of patients with no-HS.

Table 2.   Type and incidence of initial precipitating injuries (IPIs) predating onset of habitual seizure in HS (classical and atypical) groups
IPI typeNo IPI (%)FS or CFC (%)Other seizure IPI (%)Encephalitis, meningitis, postvaccination (%)Head Injury (%)Other IPI type (%)
  1. n, number of cases included in each group for which information regarding IPI available. FS, Febrile seizure; CFC, complex febrile convulsion; HS, hippocampal sclerosis; CHS, classical HS; CA1p, CA1 predominant HS; THS, total HS. No statistical differences were seen between groups regarding IPI type (p = 0.43).

No HS (n = 17)533500120
Indeterminate HS (n = 35)264631168
CA1p (n = 8)12.562.512.50012.5
End folium sclerosis (n = 53)08000200
CHS (n = 58)22.550953.510
THS (n = 39)1559101033

There were significant differences in the extent of GCD between HS groups (GCD max, p < 0.01 and GCD mean, p < 0.001), with least dispersion in CA1p and no-HS groups and greatest in CHS (Table 1). There was an inverse correlation between GCD and CA4 neuronal density (p < 0.001), but not CA1 neuronal density (p = 0.2, Pearson correlation). There was no significant correlation between GCD measures and outcome.

Discussion

  1. Top of page
  2. Summary
  3. Methods and Materials
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure
  8. References

The identification of reliable prognostic factors that could predict outcome following temporal lobe surgery for hippocampal sclerosis is desirable, as up to one-third of patients do not become seizure-free (Wieser & Hane, 2003; Aull-Watschinger et al., 2008). Studies to date have largely focused on clinical parameters, for example, preoperative seizure history and electroencephalography (EEG) findings (Spencer & Huh, 2008). Evaluation of the resected tissues may provide additional information. The less common, atypical (or “nonclassical”) patterns of hippocampal damage have long been recognized in neuropathology studies, from both postmortem epilepsy series (Margerison & Corsellis, 1966), including patients with a variety of epilepsy syndromes (Thom et al., 2009) and in surgical studies from patients with TLE (Bruton et al., 1988). Different patterns of hippocampal neuronal loss may be a manifestation of distinct pathways of epileptogenesis, timing of epilepsy onset, IPI, or other factors: HS is unlikely to be a single condition. There is evidence to suggest that atypical HS patterns may be predictive of poorer outcome (Sagar & Oxbury, 1987; de Lanerolle et al., 2003; Blumcke et al., 2007). It is likely in the future that atypical patterns of HS will be diagnosed preoperatively with more advanced magnetic resonance imaging (MRI) sequences (Eriksson et al., 2008; Mueller et al., 2009), which could influence planning of surgery. Therefore, verification of data regarding any influence of atypical patterns of HS on outcome is desirable.

We, therefore, applied a quantitative analytical method, modified from a recently published study delineating HS patterns (Blumcke et al., 2007), in an attempt to replicate those findings in our surgical series. This method utilizes NeuN staining and quantitative image analysis on single hippocampal sections. Classification of HS patterns is made through comparison to reference values from normal controls. There is a lack of suitable surgical normal hippocampal material, and we, therefore, used postmortem sections, as in Blumcke’s series. We aimed to select controls from young adults, but nevertheless the age range was older than surgical HS cases. NeuN staining is less efficient in postmortem tissue and requires a modified pretreatment regimen (Gill et al., 2005), but it is preferable to H&E for the detection of all neuronal cell types (Wolf et al., 1996). Nevertheless, because of the older ages of controls and differing immunohistochemistry methods, we cannot exclude the possibility that our control values underestimate hippocampal neuronal densities, which has bearing on the HS grouping.

In Blumcke’s study, patterns of MTS (mesial temporal sclerosis) types 1a, 1b, 2, and 3 were defined, termed CHS, THS, CA1p, and EFS groups in our series (Blumcke et al., 2007). We identified nonclassical patterns of HS in 30% of cases overall : EFS was identified in 3% and CA1p pattern was identified in 6%, comparable to 4% and 6–8%, respectively, in previous series (Bruton et al., 1988; de Lanerolle et al., 2003, Blumcke et al., 2007). We showed poorer outcomes for patients with CA1p HS, but noted better outcomes for EFS, compared to previous studies (Table 3) (Van Paesschen et al., 1997; Blumcke et al., 2007). Poorer outcomes were confirmed where quantitative evaluation revealed no significant neuronal loss (no-HS group), with 44% of patients seizure-free compared to 44% and 58% in previous large series (Table 3) (de Lanerolle et al., 2003; Blumcke et al., 2007). Differences in outcomes reported between series (Table 3) may reflect the fact that atypical HS patterns, particularly EFS, represent a very small proportion of cases and length of follow-up varies between series. The mean follow-up period was the longest in our series, and we used the ILAE classification scheme. More importantly, all patients in our study had the same operative procedure and we excluded cases in which a second potentially epileptogenic (dual) pathology was present in the resection specimen, for example, tumors, old traumatic scars, or focal cortical dysplasia. Such pathologies, included in previous series (de Lanerolle et al., 2003; Blumcke et al., 2007), could imply a different cause for the HS and independently influence postoperative outcome.

Table 3.   Comparisons of main findings in previously reported surgical studies comparing atypical patterns of hippocampal sclerosis and epilepsy history
Pattern of HSSagar and Oxbury (1987)Bruton et al. (1988)Davies et al. (1996)Van Paesschen et al. (1997)de Lanerolle et al. (2003)Blumcke et al. (2007)Present series
  1. Age onset, age of onset of habitual seizures; IPI, initial precipitating injury; FS, febrile seizures; SF, seizure free; Outcome, seizure control following surgery (using either Engel, ILAE grading (Wieser et al., 2001).

  2. aSeizure pattern greatly improved following surgery); MTS, mesial temporal sclerosis.

  3. bHabitual seizure onset age calculated by adding age at initial event and latency period.

  4. cIn these series CA1p and EFS HS grouped together as one category/grade.

CHS  No. of cases  Age onset  IPI (FS)  OutcomeCHS  12  9.1 years  No data  No dataCHS  61  5.19 years  48% (FS)  37% (SF)aCHS  59  7.4 years  No data  68% (SF)CHS  52  7 years  62% (FS)  81% (SF)CHS  72  4.4 years  No data  84.5% (SF)CHS/MTS type 1  33  10.3 yearsb  76% (FS)  72% (SF)CHS  60  9.3 years  50% (FS)  69% (SF)
EFS  No. of cases  Age onset  IPI type (FS)  OutcomeEFS+CA1pc  10  13.2 years  No data  No dataEFS  4  18.5 years  0% (FS)  50% (SF)EFS+CA1pc  8  12.4 years  No data  66.7% (SF)EFS  4  15.5 years  0%  25% (SF)EFS  No casesEFS/MTS type 3  7  14 years  0%  28% (SF)EFS  5  18.2 years  80% (FS)  100% (SF)
CA1P  No. of cases  Age onset  IPI type (FS)  OutcomeCA1P  No casesCA1P  No casesCA1P  9  7.9 years  No data  77.8% (SF)CA1P/MTS type 2  10  15 yearsb  75% (FS)  66.7 (SF)CA1P  9  5.9 years  62.5% (FS)  33% (SF)
No HS  No. of cases  Age onset  IPI type (FS)  OutcomeNo HS  10  11 years  No data  No dataNo HS  38  15.37 years  2.6% (FS)  20% (SF)aNo HS  31  19.9 years  No data  42.3% (SF)No HS  No casesNo HS  18  9.4 years  No data  44% (SF)No HS  34  18.4 yearsb  0% (FS)  58.6% (SF)No HS  17 cases  12.4 years  35% (FS)  44%
Significance reported between groupsp < 0.02 (age onset)No statistics reported between HS groups Follow-up period, 8 years (mean).p < 0.001 (age onset) p < 0.003 (IPI type, FS) p < 0.01 (outcome)p < 0.02(age onset) p < 0.03 (IPI type, FS) p < 0.04 (outcome) Follow-up period 1 year (Engel)p < 0.05 (age onset) Follow-up period 1 year (Engel)p < 0.01 (age onset) p < 0.001 (IPI type, FS) p < 0.04 (outcome) Follow-up period 1 year (Engel)p < 0.01 (age onset) p < 0.05 (outcome) Follow-up period 2 years (ILAE)

Clinicopathologic correlations have also highlighted differences in seizure histories with different patterns of HS. Later ages of onset of habitual seizures have been reported in atypical HS (Table 3). A later onset in EFS cases was noted in our series compared to other HS types, as also shown in previous series (Sagar & Oxbury, 1987; Van Paesschen et al., 1997; Blumcke et al., 2007). It has been proposed that atypical patterns of HS may be a manifestation of variable rates of hippocampal neuronal maturation, influencing vulnerability and cell-protective mechanisms in different hippocampal subfields (Van Paesschen et al., 1997; Blumcke et al., 2007). Regarding the initial insult, we did not find any significant differences in the age or type of IPI between the HS patterns. Of note, in contrast to previous studies, a history of febrile seizures was a frequent occurrence in EFS cases (80%), whereas, in keeping with previous studies, this was less common in the no-HS group (Blumcke et al., 2007).

Quantitative analysis is a time-consuming process but may provide objective criteria for grouping HS cases over semiquantitative grading systems such as the Wyler score or similar schemes (Davies et al., 1996; Thom et al., 2002), as well as additional information. This is exemplified by our identification of a group with indeterminate patterns of sclerosis forming 21% of the series. In 18 of these cases, significant neuronal loss involved both CA3 and CA4. On qualitative inspection alone such cases would be allocated to the “best-fit” HS category or grade, for example EFS. These indeterminate HS patterns were associated with poorer seizure-free outcomes compared to classical HS, motivating their identification and further delineation using quantitative methods.

Granule cell dispersion (GCD) is a common and as yet unexplained phenomenon of uncertain clinical significance (Fahrner et al., 2007), occurring in 40% of epilepsy HS specimens overall (Thom et al., 2002; Wieser, 2004). The criteria for determining GCD vary from qualitative impressions of increased thickness of >10 cells in the GCL (Wieser, 2004; Blumcke et al., 2007) to measurements using image analysis. Mean thicknesses of 140–250 μm have been reported in HS (Mathern et al., 1997; Thom et al., 2002; Fahrner et al., 2007), with values of 120 μm or more considered as indicative of GCD (El Bahh et al., 1999). In our series, mean GCD measurements in HS groups ranged from 124–213 μm compared to the control group mean of 79 μm. Our method did not address differences between the limbs of the dentate gyrus but evaluated the overall mean and maximal GCD in each case. There was a correlation between the extent of GCD and the pattern of HS, with lowest GCD in CA1p and no-HS groups, as noted in previous studies (de Lanerolle et al., 2003; Blumcke et al., 2007). There was also a correlation between GCD and CA4 (but not CA1) neuronal loss. This would support the theory that GCD is associated with the presence of hilar neuronal damage (Mathern et al., 1997; Thom et al., 2002). In keeping with a recent study (Blumcke et al., 2009) our findings suggests that GCD does not appear to have any influence on outcome following surgery.

There are both limitations and advantages to the methodology employed in this study. The quantitative analysis was carried out on only one representative section of hippocampus. There is a possibility that patterns of sclerosis and GCD could vary along the length of the hippocampus in some cases (Thom et al., 2002). Qualitative assessment is typically based on H&E and GFAP sections. NeuN sections could provide additional evidence for subfield damage. This could explain the lack of exact correlation between quantitative and qualitative findings in the study. In addition, we did not analyze cell densities in the granule cell layer and subiculum, areas usually considered in the qualitative assessment of HS. As with any histologic study, there is an assumption that tissue volume changes (which could influence cell density measurements) occur uniformly between cases. In addition, we did not measure neuronal size or the area of each subfield, which could also influence cell density measurements. Clinical observations are dependent on the accuracy of the case records, and as in all clinicopathologic studies there may be inaccuracies regarding recollection of IPI events and seizure history. One advantage of our study is that all patients were treated at the same center using a standardized surgical technique (anterior temporal lobectomy). We also excluded patients for whom a second lesional pathology was identified, allowing a clearer assessment of the effects of hippocampal pathology alone on outcome. Our longer follow-up periods of 2 years in the majority of patients compared to other series, and the use of the ILAE classification system (Wieser et al., 2001) has allowed a more stringent analysis of the long term effects of HS type on outcome. In conclusion, our study highlights the potential additional value of quantitative analysis of HS in the prediction of surgical outcome.

Acknowledgments

  1. Top of page
  2. Summary
  3. Methods and Materials
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure
  8. References

We are very grateful to Jane de Tisi for her administrative help, and special thanks to Catherine Godfraind, Professor of Neuropathology University of Louvain and Linda Kilford of the Queen Square Brain Bank for their help with obtaining control tissues. We are also grateful to Denise Watson for her help with section preparation. Lillian Martinian is supported by the MRC (Grant G79059). This work was undertaken at UCLH/UCL, which received a proportion of funding from the Department of Health’s NIHR Biomedical Research Centres funding scheme. LOC was supported by CAPES, Brazil.

Disclosure

  1. Top of page
  2. Summary
  3. Methods and Materials
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure
  8. References

We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines. None of the authors had any conflict of interest to disclose.

References

  1. Top of page
  2. Summary
  3. Methods and Materials
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure
  8. References