Methods

STUDY POPULATION

Forty six patients (18 men, 28 women; median age 31, range 17-51 years) with intractable temporal lobe epilepsy who underwent anterior temporal lobe resection were included in the study. The body of the hippocampus of all of these patients was available for histopathological examination. Qualitative microscopical assessment of 41 hippocampi showed hippocampal sclerosis and five hippocampi showed mild gliotic changes confined to the end folium.7 Six control hippocampi of people (three men, three women; median age 31, range 14-52 years) who had died from non-neurological causes were obtained at necropsy. None of the control hippocampi showed hypoxic or other neuropathological changes.

CLINICAL EVALUATION

Age at onset of habitual epilepsy, duration of epilepsy, a history of febrile convulsions and meningoencephalitis, age at the time of these events, family history of febrile convulsions and epilepsy, seizure types and description, average frequency of each seizure type during the year preceding the operation, and total number of secondary generalised seizures in their lifetime were established. In those who had a postoperative follow up of at least one year, outcome was rated as 1, no seizures or auras only, 2 >90% reduction in seizures, 3 >50% reduction in seizures, or 4 no improvement. Clinical evaluation of patients, quantitative hippocampal MRI, and quantitative neuropathological studies were performed by the same observer.

QUANTITATIVE MRI

Presurgically, MRI based hippocampal T2 (HCT2) and volumes (HCV) were determined as described previously.8 Mean control (SD) HCT2 was 102.4 (2.8) ms, and mean control HCV corrected for intracranial volume (SD) was 5180 (416) mm³.

NEUROPATHOLOGY

Hippocampi were removed en bloc during surgery and were fixed immediately in formalin for one week. After fixation each specimen was sliced into 3 mm thick tissue blocks. The plane of sectioning was perpendicular to the long axis of the hippocampus. Sections of each paraffin embedded block, 5 to 7 μm thick, were stained with haematoxylin and eosin, luxol fast blue, and glial fibrillary acidic protein for neuropathological assessment. For quantification of corpora amylacea, one 20 μm thick luxol fast blue-cresyl violet stained section of the body of the hippocampus with the characteristic C shaped appearance of the cell layers, corresponding to the HCT2 map, was selected. A direct three dimensional counting method was used.9 We used a Zeiss microscope fitted with a Zeiss drawing tube, a digital length gauge (Heidenhain) with a sensitivity of 0.5 μm that was firmly attached to the stage, and an oil immersion lens with a magnification of ×100 and numeric aperture of 1.3 with a depth of field of 0.22 μm. For measurements of corpora amylacea density, a counting box of 50×50×10 μm was used, unless the specimens had high densities of corpora amylacea, when the counting box was 20×20×10 μm. We followed the counting rules recommended by Williams and Rakic.9 Corpora amylacea completely inside the counting box were counted, and those completely outside the counting box were not. Corpora amylacea that touched the forbidden planes—that is, bottom, front, and left side of the counting box—were excluded, and those that touched the top, right side, and rear of the counting box were counted, provided that they did not touch any of the forbidden planes. Corpora amylacea were defined as globular bodies which stained blue with luxol fast blue. For the definition of the neuron containing layers of the hippocampus, we used the descriptions of Lorente de Nó,10 Duvernoy,11 and Amaral and Insausti.12 Corpora amylacea densities were determined for the pyramidal cell layer of CA1, CA2, CA3, hilus, and the granular cell layer of the dentate gyrus (GCDG). Each hippocampal subregion was counted using 50 counting boxes with a random and systematic sampling strategy. Using similar methodology, neuronal cell densities were determined in the same hippocampal subregions, as reported previously.6 Statistical analysis was performed using SPSS for Windows, release 6 (SPSS Inc, Chicago, IL, USA). Spearman’s correlation coefficient (r) was used for correlation of corpora amylacea and neuronal cell densities, a χ² test for comparison of categorical variables, and a Mann-WhitneyU test for comparison of continuous variables in patients with and without corpora amylacea. This study was approved by the ethics committee of the National Hospital for Neurology and Neurosurgery.