In the present study, greater improvement of depressive symptoms (reductions in HDRS score) with chronic citalopram treatment was associated with lower pre-treatment cerebral glucose metabolism and greater reductions in cerebral glucose metabolism after acute, intravenous citalopram administration. At baseline, higher glucose metabolism in the rostral anterior cingulate (BA 24, 32), frontal and parietal cortices, striatum (caudate, putamen) and limbic/paralimbic regions (insula, parahippocampal gyrus) was associated with less improvement of depressive symptoms with citalopram treatment. This finding is consistent with observations of cortical hypermetabolism in geriatric depressed patients relative to controls and of correlations between higher metabolism and greater severity of depression and anxiety symptoms in the patients prior to treatment (13
). Several regions including the left medial frontal gyrus, right superior temporal gyrus, left fusiform gyrus and bilateral cerebellum showed the opposite association (greater improvement associated with higher metabolism), which may represent a compensatory response in these mainly sensory and motor regions for change in other brain regions.
The reductions in cerebral metabolism after acute citalopram administration that were correlated with improvement of depressive symptoms occurred in many of the same regions in which the baseline associations were observed. The notable differences between the baseline and acute citalopram analyses were that the striatal regions showed correlations with baseline metabolism only, whereas more extensive changes in posterior cortical regions were observed in the acute citalopram condition only (e.g. right supramarginal and fusiform gyri, left lingual gyrus, left superior and middle temporal gyri). The correlations between less depressive symptom improvement and higher striatal metabolism in the baseline condition and not the acute citalopram condition suggest that the underlying mechanism may be another neurotransmitter aside from serotonin. Other neurochemical mechanisms including increased glutamate concentrations or structural pathology in fronto-striatal circuitry that might result in an increase by disinhibition of striatal metabolism may be involved (11
). The left middle frontal gyrus was the only region that showed the opposite association (increased metabolism was associated with greater improvement of depressive symptoms. This was the only region that showed decreased metabolism in the geriatric depressed patients compared to controls which may explain this opposite finding (13
). With respect to gender differences, the male depressed patients showed significant greater correlations between metabolic responses and clinical improvement than females even though the magnitude of clinical improvement did not differ significantly between groups. The greater association between clinical and metabolic responses in males is consistent with reports of greater serotonin metabolism in males than females (23
) and may be associated with the greater vulnerability of females than males to depression (24
In comparing the correlations with treatment response for the acute citalopram study to the earlier study of correlations between the metabolic effects of TSD and paroxetine response (1
), similar anterior cortical regions show correlations in both studies. The acute citalopram data show a more extensive brain network of correlations including more posterior cortical regions. This observation may be explained by a greater neurochemical effect of acute citalopram compared to TSD or possibly differences in patient characteristics, although the samples are similar in such variables as magnitude and rate of treatment response. Neurochemical brain imaging studies of the acute effects of intravenous citalopram have shown significant serotonin transporter occupancy in striatum, thalamus, brainstem, amygdala and hippocampus, as well as increases in striatal dopamine concentrations (25
). Thus, both primary and secondary neurochemical effects of citalopram have been observed in the same time frame as the cerebral metabolic effects observed in the present study. Given the regional distribution of the correlations, the alterations in cerebral metabolism may reflect the effects of serotonin transporter occupancy on cortico-cortico circuits that are likely to be mediated through a glutaminergic mechanism (27
). In addition, many of the regions that are “hypermetabolic” at baseline and are affected by citalopram treatment are regions that comprise the “default network” and that demonstrate beta-amyloid deposition in demented and non-demented elderly (29
). As increased glutamate activity is observed in amyloid transgenic mouse models and serotonin inhibits cortical glutamate (30
), a secondary consequence of beta-amyloid deposition and decreased serotonin functional integrity could be glutamate hyperactivity which would increase glucose metabolism (28
). Citalopram treatment may decrease metabolism by decreasing glutamate concentrations (31
). Thus, correlations in the resting state and after acute citalopram administration provide different functional neuroanatomic and mechanistic information associated with depressive symptom improvement.
The majority of studies that have evaluated baseline or acute cerebral metabolism or neuroreceptor measures relative to clinical treatment outcome have been performed in mid-life depressed patients. There is some evidence, consistent with findings of the present study, that baseline hypermetabolism of between the pregenual and subgenual cingulate cortices (BA 24/32) predicts worse treatment response to venlafaxine or cognitive behavioral therapy (CBT; 32). Lower midbrain metabolism was also associated with better antidepressant treatment response (33
). Other studies of SSRI treatment and TSD in mid-life depressed patients have reported opposite findings in rostral anterior cingulate cortex and frontal cortex (34
). As described, the findings in older depressed patients may be a secondary consequence of neuropathological processes, in contrast to findings in some studies of mid-life depressed patients (34
). Magnetic resonance spectroscopy (MRS) studies suggest that bioenergetic abnormalities related to mitochondrial dysfunction may be associated with treatment response. These studies have shown that 1) lower basal ganglia beta nucleoside triphosphate and purine intensities (in females) was associated with better treatment response to fluoxetine (36
) and 2) higher baseline phosphocreatine was associated with better treatment response to triiodothyronine (T3) augmentation (37
). Serotonin imaging studies have shown that increased 5-HT1A binding in cortical and limbic regions and the raphe nuclei and lower serotonin transporter (5HTT) binding in the anterior cingulate, amygdala and midbrain were associated with poorer SSRI treatment response (38
). As increased 5-HT1A binding may represent an upregulation due to decreased serotonin concentrations and the lower transporter binding may suggest a loss of serotonin projections, these observation suggest serotonin hypofunction in non-responders. These observations are consistent with the findings of the present study that suggest a blunted metabolic response to acute citalopram is associated with a poorer antidepressant response.
Several issues should be considered in the interpretation of the data from the present study. As described in the methods section, the acute citalopram administration was performed in a fixed order. The chronic citalopram treatment phase was open label and a placebo treated group was not included. It is important to note that the rate of treatment non-response to citalopram in this study (25%) was similar to that of placebo controlled trials (6
). Another important consideration is that the HDRS measures the core symptoms of depressed mood, in addition to the vegetative signs of depression, so the correlations represent metabolic alterations associated with the net effect on different aspects of depressive symptomatology.
In summary, lower pretreatment cerebral glucose metabolism in the rostral anterior cingulate, prefrontal cortex, striatum and limbic/paralimbic regions was associated with greater improvement of depressive symptoms in geriatric depressed patients. Furthermore, greater decreases in cerebral glucose metabolism following administration of a serotonergic challenge agent were associated with a greater antidepressant effect of citalopram. The regions affected uniquely in the acute citalopram condition compared to baseline metabolism were temporal, parietal and occipital cortical regions. The anterior cortical findings are consistent with the prior study that showed an association between the cerebral metabolic effects of TSD and the antidepressant response (1
). The regional pattern of the acute citalopram effects in cortical and limbic regions includes the targets of serotonergic projections which suggest that the changes in metabolism reflect the functional integrity of the serotonin system or serotonin modulation of glutamate concentrations (40
). These regional cerebral metabolic findings are consistent with the results of other studies in geriatric depression that have shown correlations between poorer treatment response and white matter connectivity in similar cortical and limbic pathways (11
). The results of the present study suggest that the dynamic response of the brain to an acute increase in serotonin, the cerebral metabolic response to acute citalopram, is indicative of the capacity of the brain to respond to a course of chronic citalopram treatment. Future studies should evaluate the extent to which the acute metabolic and neurochemical changes associated with other pharmacologic classes of antidepressants (e.g. selective noradrenergic reuptake inhibitors), somatic treatments (electroconvulsive therapy, transcranial magnetic stimulation) and psychotherapy are associated with improvement in depressive symptoms to determine whether the early metabolic neurochemical changes represent biomarkers of treatment response. While the integration of neuroimaging studies into clinical trials is logistically challenging, such studies are critical to developing neuroimaging biomarkers of treatment response in psychiatric disorders.