DG of aged rats 1d and 3d after hepatectomy, but not at cingulate cortex neurons of aged rats and at the neurons of CA1, DG and cingulate cortex of young adult rats. These data showed that the loss of spines of neurons at CA1 and DG was closely associated with postoperative memory impairment of aged rats; hippocampus was the main target impaired by surgery. This was consistent with our previous research of aged patents with surgery. In aged patients, small hippocampal volume could be an independent risk predictor of POCD. In addition, Bloss et al found that dendritic spines of prefrontal cortex neurons of aged rat were remarkably stable to the stress. Thus it was possible that surgery didn’t affect the dendritic spine density of neurons at cingulate cortex of aged rats. The difference of frontal cortex involved POCD between human and rats reported in this study might be due to the species difference. In addition, anesthesia or surgery, like other stress factors, may provide dual effects on neuroprotection and neurotoxicity. Minor stresses may provide neuroprotection. Detrimental stresses may provide neurotoxicity. The stress size of anesthesia or surgery was also the possible reason for the difference of frontal cortex involved POCD between human and rats reported in this study. Age is the main risk factor of POCD. POCD is usually detected in aged patients with surgery, but not in young adult patients. Low cognitive reservation is thought to be the reason for occurrence of POCD at aged patients. However the question is how age influences POCD of young adult and aged animals. Thus we first detected the spine density of neurons at young adult and aged rats after surgery. Our data showed that in contrast to the significant loss of dendritic spines of neurons at CA1 and DG of aged rats after surgery, there were no obvious changes at spine densities of dendrites of neurons at CA1, DG and cingulate cortex of young adult rats. These suggested that surgery Epimedoside-A induced more obvious impairment of neurons at aged rats than that of young adult rats. Neuroinflammation was closely associated with POCD. So we also detected the activation of microglia and the expressions of TNF-a and IL-1b after surgery. Corresponding to the loss of dendritic spines of hippocampal neurons, microglia was activated and levels of TNF-a and IL-1b were up-regulated at hippocampus of aged rats after surgery. In contrast, activation of microglia and Gomisin-D increase of TNF-a and IL-1b were not detected in young adult rats after surgery. These data showed that surgery induced strong neuroinflammation at the hippocampus of aged rats, but not at young adult rats. Similar results were also reported by Cao. They found that surgery induced more durable and stronger inflammation response in aged rats, compared with adult rats. Previous studies showed that intra-hippocampal administration of IL-1b impaired contextual fear memory of rats. Sustained elevation of hippocampal IL1b levels also produced marked impairments in spatial memory. In addition, over-expressing TNF-a in the brain impaired leaning of adult mice. Intrahippocampal administration of TNF-a impaired hippocampaldependent working memory. Blocking the signals of TNF-a and IL-1b effectively decreased the cognitive function impairment induced by surgery. These information showed that increase of TNF-a and IL-1b was detrimental in learning and memory. Based on the above information, we thought that strong neuroinflammation was possible mechanism for significant loss of dendritic spines of 
hippocampal neurons.