Quantification of subfield pathology in hippocampal sclerosis: A systematic review and meta-analysis

Background: The utility of MRI-based hippocampal subfield volumetry as a diagnostic test for hippocampal sclerosis (HS) is based on the hypothesis that specific hippocampal subfields are differentially affected in HS. While qualitative studies suggest selective involvement of certain hippocampal subfields in this condition, whether quantifiable differences exist remains unclear. Neuronal density measurement is the most widely used technique for measuring subfield pathological change in HS. Therefore, a systematic review and meta-analysis of studies reporting neuronal densities in temporal lobe epilepsy was performed in order to quantify subfield pathology in hippocampal sclerosis. Methods: Studies were identified by searching the Med line and Embase databases using the search terms: cell count, hippocampus, and epilepsy. Of the 192 studies identified by the literature search, seven met all inclusion and exclusion criteria. Random effects meta-analyses were performed, comparing: (i) neuronal densities in control (n = 121) versus HS (n = 371) groups for subfields CA1-4; and (ii) amount of neuronal loss in HS between subfields CA1-4. Results: Statistically significant neuronal loss was observed comparing HS to control groups in all subfields CA1-4 (p < 0.001 for all comparisons). Significantly greater neuronal loss was demonstrated in HS comparing CA1 versus CA2 (p < 0.001), CA3 (p = 0.005), and CA4 (p = 0.003). Greater pyramidal cell loss was also demonstrated in CA3 relative to the CA2 subfield (p = 0.003). No significant differences were identified comparing CA2 and CA4 (p = 0.39); or comparing CA3 and CA4 (p = 0.64). Conclusions: HS is characterized by pathology in all hippocampal subfields. Quantifiable differences exist in the involvement of specific hippocampal subfields in HS. Neuronal loss is greatest in CA1, intermediate in CA3 and CA4, and least in CA2. Further studies are required to determine if this pattern can be detected using in vivo MRI. (C) 2014 Elsevier B.V. All rights reserved.