Failure of hippocampal deactivation during loss events in treatment-resistant depression

In 1991, Deakin and Graeff published a series of hypotheses on serotonergic activity in depression. Johnston et al. use fMRI to test these predictions in subjects with treatment-resistant depression performing an instrumental learning task. The results largely support the hypotheses, particularly the observed correlation between hippocampal overactivity during loss events, and illness severity.In 1991, Deakin and Graeff published a series of hypotheses on serotonergic activity in depression. Johnston et al. use fMRI to test these predictions in subjects with treatment-resistant depression performing an instrumental learning task. The results largely support the hypotheses, particularly the observed correlation between hippocampal overactivity during loss events, and illness severity.Major depressive disorder is characterized by anhedonia, cognitive biases, ruminations, hopelessness and increased anxiety. Blunted responses to rewards have been reported in a number of recent neuroimaging and behavioural studies of major depressive disorder. In contrast, neural responses to aversive events remain an under-studied area. While selective serotonergic reuptake inhibitors are often effective in treating major depressive disorder, their mechanism of action remains unclear. Following a series of animal model investigations of depressive illness and serotonergic function, Deakin and Graeff predicted that brain activity in patients with major depressive disorder is associated with an overactive dorsal raphe nucleus with overactive projections to the amygdala, periaqueductal grey and striatum, and an underactive median raphe nucleus with underactive projections to the hippocampus. Here we describe an instrumental loss-avoidance and win-gain reinforcement learning functional magnetic resonance imaging study with 40 patients with highly treatment-resistant major depressive disorder and never-depressed controls. The dorsal raphe nucleus/ periaqueductal grey region of the midbrain and hippocampus were found to be overactive in major depressive disorder during unsuccessful loss-avoidance although the median raphe nucleus was not found to be underactive. Hippocampal overactivity was due to a failure to deactivate during loss events in comparison to controls, and hippocampal over-activity correlated with depression severity, self-report 'hopelessness' and anxiety. Deakin and Graeff argued that the median raphe nucleus normally acts to inhibit consolidation of aversive memories via the hippocampus and this system is underactive in major depressive disorder, facilitating the development of ruminations, while the dorsal raphe nucleus system is engaged by distal cues predictive of threats and is overactive in major depressive disorder. During win events the striatum was underactive in major depressive disorder. We tested individual patient consistency of these findings using within-study replication. Abnormal hippocampal activity correctly predicted individual patient diagnostic status in 97% (sensitivity 95%, specificity 100%) of subjects, and abnormal striatal activity predicted diagnostic status in 84% (sensitivity 79%, specificity 89%) of subjects. We conclude that the neuroimaging findings were largely consistent with Deaken and Graeff's predictions, abnormally increased hippocampal activity during loss events was an especially consistent abnormality, and brainstem serotonergic nuclei merit further study in depressive illness.