Different patterns of epileptiform-like activity are generated in the sclerotic hippocampus from patients with drug-resistant temporal lobe epilepsy

Abstract

Human hippocampal slice preparations from patients with temporal lobe epilepsy (TLE) associated with hippocampal sclerosis (HS) are excellent material for the characterization of epileptiform-like activity. However, it is still unknown if hippocampal regions as cornu Ammonis (CA) 1, CA3 and CA4, generate population epileptiform-like activity. Here, we investigated epileptiform activities of the subiculum, CA1, CA2, CA3, CA4 (induced by elevation of extracellular potassium concentration) and the dentate gyrus (induced with hilar stimulation and elevation of potassium concentration) from sclerotic hippocampi of patients with drug-resistant TLE. Five types of epileptiform-like activity were observed: interictal-like events; periodic ictal spiking; seizure-like events; spreading depression-like events; tonic seizure-like events and no activity. Different susceptibilities to generate epileptiform activity among hippocampal regions were observed; the dentate gyrus was the most susceptible region followed by the subiculum, CA4, CA1, CA2 and CA3. The incidence of epileptiform activity pattern was associated with specific regions of the hippocampal formation. Moreover, it was observed that each region of the hippocampal formation exhibits frequency-specific ranges in each subfield of the sclerotic human tissue. In conclusion, this study demonstrates that epileptiform-like activity may be induced in different regions of the hippocampal formation, including regions that are severely affected by neuronal loss.

Figure 1

Epileptiform-like activity in hippocampal formation. Representative recordings of the different patterns of epileptiform-like activity in different regions of hippocampal formation. (A) Interictal-like events. (B) Periodic ictal spiking. (C) Seizure-like event. (D) Spreading depression-like events. (E) Tonic seizure-like events. (F) Non-epileptiform activity. Isoelectric recordings indicate that non-epileptiform activity was observed. Calibration bars for amplitude and time are given in each recording on the right.

Figure 2

Distribution of epileptiform activity. (A) Conditional probability distribution of generated specific epileptiform-like activity within the hippocampal subfields P (epileptiform activity|hippocampal area). (B) Conditional probability distribution of specific hippocampal formation area to generate epileptiform activity P (hippocampal area|epileptiform activity). Bar graphs represent the mean probability with the corresponding standard deviation. Circular graphs represent the absolute frequencies. SLE: Seizure-like events; T-SLE: Tonic seizure-like events; SD: Spreading depression-like events; PIS: Periodic ictal spiking; II: interictal-like events.

Figure 3

Histopathological patterns of hippocampal sclerosis. (A–D) NeuN immunostaining of neuronal nuclei. (A) No hippocampal sclerosis in control specimen obtained from autopsy without neurological disease. (B) ILAE hippocampal sclerosis type 1, note the marked neuronal loss in both the CA1 and CA4 subfields. (C) ILAE hippocampal sclerosis type 2, predominant cell loss in the CA1 area. (D) ILAE hippocampal sclerosis type 3, most restrict neuronal loss in the CA4 sector. Among all hippocampal sclerosis types, the histological pattern is variable in the dentate gyrus, wherein the granular cell loss and dispersion is predominant. Hippocampal subfields and the dentate gyrus were observed with a magnification of 20×. Arrows indicate regions with severe neuronal loss. (E) Measurement of neuronal density in hippocampal subfields, neuronal cell counting was transformed into z-score using values from control hippocampi. Hippocampal sclerosis sectors were determined when cell neuronal counting was <than −2 in z-score values. For the dentate gyrus, granular cell dispersion was considered when width reached >2 in z-score values.

Figure 4

Distribution of epileptiform activity by ILAE HS type. (A) Conditional probability distribution of the different HS types to generate specific patterns of epileptiform activity P (HS type|epileptiform activity). (B) Conditional probability distribution of the different HS types to present epileptiform activity susceptibility in the different hippocampal subfields P (HS Type|hippocampal area). Bars graphs represent the mean probability with the corresponding standard deviation. Circular graphs represent the absolute frequencies. SLE: Seizure-like events; T-SLE: Tonic seizure-like events; SD: Spreading depression-like events; PIS: Periodic ictal spiking; II: interictal-like events.

Figure 5

Electrophysiological signal analysis. Electrophysiological analysis of one illustrative patient (case one) is showed in (A–D). Analysis of the other four patients also selected in this study are found in the supplementary information. (A) Averaged waveform of epileptiform patterns in each hippocampal subfield was obtained using the spike sorting-like procedure. The waveform in color represents the average of trial recordings, with 95% confidence intervals (gray background). (B) Principal component analysis showing different clusters among hippocampal subfields. (C) Power frequency of the different hippocampal subfields; note that each subfield exhibited different power-frequency ranges. The average power frequency is represented by the continuous line in color with their respective confidence interval of α = 0.05. (D) Principal component analysis applied on the frequency domain; observe the different trajectories and critical points in each hippocampal subfield. (E–G) shows analysis of all five tissue samples. (E) Maximal frequencies displayed by the epileptiform patterns in different hippocampal subfield. Bar graphs represent the means and standard deviation. (F) Cluster centroid distances in each hippocampal subfield between original PCA (red circles) and PCA after filtering recordings using the critical points in each trajectory (green circles) and with minimal and maximal frequencies (blue circles). The distance centroid is represented by the confidence interval mean and its standard deviation. (G) Percentage of explained variances of PCA with filtered and unfiltered data. Bar graphs represent the mean and standard deviation.

Figure 6

Methods for electrophysiological signal analysis. (A) By spike sorting-like procedure, epileptiform patterns were selected and averaged by confidence intervals (α = 0.05). (B) After obtaining the waveform matrices, PCA was applied to the global matrix $\mathrm{X}\left(\tau \right)$. (C) For frequency domain analysis, each epileptiform pattern was processed by the wavelet technique, an average of sample matrices and wavelet-matrices for each subfield was obtained. The power spectrum was calculated from the global time-scale matrix, and PCA was applied.