In women with anorexia nervosa (AN), an enlarged ventricular volume and dilated sulci have been consistently found (Dolan et al., 1988, Krieg et al., 1988, Krieg et al., 1989, Kingston et al., 1996, Swayze et al., 1996 and Swayze et al., 2003). Reductions in both white and gray matter have also been observed (Katzman et al., 1996 and Swayze et al., 2003). Brain metabolism studies parallel the structural findings, showing global reductions in brain metabolism (Delvenne et al., 1995) and perfusion (Krieg et al., 1989). Regional perfusion studies of children and adolescents with AN demonstrated unilateral reduced perfusion in the temporal lobe (in which the hippocampus is located) in the majority of those studied (Gordon et al., 1997 and Chowdhury et al., 2003). In contrast, one unreplicated study reported relative hyperperfusion of the thalamus and the amygdala–hippocampal complex in adults with active AN (Takano et al., 2001). Structural deficits tend to improve with weight gain (Dolan et al., 1988, Krieg et al., 1988, Krieg et al., 1989, Kingston et al., 1996, Swayze et al., 1996 and Golden et al., 1996), although gray and white matter volume may remain significantly reduced following recovery (Lambe et al., 1997, Katzman et al., 1997 and Swayze et al., 2003) and temporal lobe hypoperfusion may also persist (Rastam et al., 2001). Significant reductions in the volume of the hippocampal–amygdala formation have also been reported (Giordano et al., 2001). It is unclear whether these various persisting abnormalities reflect premorbid vulnerability factors for AN or scars of the illness. One possible pathogenic mechanism is hypercortisolaemia, which along with other factors related to malnutrition may contribute to the differential recovery rates of the various structural abnormalities in AN (Naruo, 2002 and Swayze et al., 2003). Cortisol is known to specifically endanger hippocampal neurones (Sapolsky et al., 1986), and is associated with both global hypometabolism (Brunetti et al., 1998) and significant hippocampal atrophy in Cushing's disease (Starkman et al., 1992).
The hippocampus is involved in learning and memory (Zola-Morgan and Squire, 1990), and discrete hippocampal lesions give rise to impaired recall with relative sparing of recognition (Baddeley et al., 2001 and Mayes et al., 2002). An association between reduced hippocampal volume and impaired verbal recall has been observed in Cushing's disease (Starkman et al., 1992), depressive disorder (Sheline et al., 1999), posttraumatic stress disorder (PTSD) (Bremner et al., 1995) and normal aging (Lupien et al., 1998). Significant deficits of learning and memory have been demonstrated in AN (Jones et al., 1991, Mathias and Kent, 1998 and Kingston et al., 1996), with some weak evidence of an association between memory impairment and structural brain abnormalities (Kingston et al., 1996).
The aim of the study was to test the hypothesis that hippocampal volume would be reduced in underweight AN and that it would be associated with impaired hippocampus-dependent cognitive function. Furthermore, hippocampal volume would be inversely related to the duration of the disorder, supporting the hypothesis that structural deficits are a scar of the illness.
