Three weeks ago I looked at neuropharmacological aspects of the neurobiology of depression. It’s time to fill in the picture with a discussion of neuroimaging research from two recent reviews. Manpreet Singh and Ian Gotlib of Stanford summarize findings from structural and functional MRI studies, and Jeffrey Meyer from the University of Toronto discusses PET, SPECT, and proton MRI studies, all of which use radiolabeled molecules to look at particular neurotransmitters or receptors. Unsurprisingly, the results are very complex and sometimes contradictory, with no clear picture emerging, but, as we know, depression is a clinical syndrome which may have multiple, possibly overlapping, biologies.
To begin with a note of caution: as Daniel Weinberger and Eugenia Radulescu recently pointed out, imaging studies in psychiatric patients, particularly structural, resting-state studies, are susceptible to numerous technical artifacts like head motion and breathing effects, as well as confounders like smoking, body weight, metabolic factors, medical cormorbidities, psychotropic drugs, alcohol use, and mental state. Weinberger and Radulescu also criticize the assumptions behind “resting state” functional MRI (fMRI) studies comparing patients’ and controls’ “default mode network” activity. They note that the brain is always active and doesn’t really have a default mode, and it is impossible to know what cognitive and emotional processes are occur in the minds of individuals unless they are assigned mental tasks (and even then tremendous individual variation is likely.) Looking at changes in individual patients’ brains in response to an external stimulus reduces these problems to some extent.
Structural MRI studies compare individuals with depression with non-depressed controls, or patients with depression during depressed and healthy phases of illness. Some have looked at non-depressed first-degree relatives or at young people at high risk for depression who have not yet become depressed. Singh and Gotlib note fairly consistent differences associated with depression in the amygdala and hippocampus. In particular, reduced hippocampal volume correlates with the number and duration of episodes of depression, and Meyer notes that this is associated with deficits in verbal memory, and that hippocampal volume loss persists even after remission of depression.
Functional MRI measures regional blood flow and can show changes in brain activity in response to emotional and cognitive stimuli. A plethora of psychological tasks has been used, and it is difficult to discern consistent patterns. One thing is clear: depression is not just a state of under-activity in the brain. Typical patterns involve over-activity is some areas, such as the amygdala and associated limbic circuits, and under-activity in others, typically regions of the prefrontal cortex. They also highlight recent research using fMRI scans of high-risk individuals to identify who is likely to develop depression; structural and functional changes have been found to precede the onset of depressive symptoms. Researchers are also developing imaging techniques to predict response to treatment. Some abnormalities appear to normalize after treatment, while others persist, which may reflect underlying traits reflecting vulnerability to depression, or the effects of depression on the brain.
Singh and Gotlib note the importance of integrating neuroimaging findings with other neurobiological research on brain networks, including the hypothalamic-pituitary axis, neurotrophins and other growth factors, the micro-RNA’s which regulate post-transcription gene expression, pro-inflammatory cytokines, and gastrointestinal signaling peptides, all of which contribute to mood alterations.
Meyer’s review is a move in that direction, looking at neuroimaging as well as postmortem studies of monoamine metabolism, serotonin and dopamine activity, and hippocampal volume. He also attempts to identify clinical markers for abnormal brain activities in these areas.
Levels of monoamine oxidase A (MAO-A), which breaks down dopamine, norepinephrine, and serotonin, are highest in the locus coeruleus, located in the pons and the principal site for synthesis of norepinephrine. Intermediate levels are found in the cortex, hippocampus, and striatum. Higher MAO-A levels are associated with depression, and high levels persisting after treatment predict relapse. Interestingly, in postpartum mood states, the rapid decrease in estrogen levels after delivery has been associated with a rapid rise in MAO-A levels. Measurement of platelet MAO in the blood, which we sometimes did back in the 1980’s, reflects levels of a different enzyme, MAO-B, and is not correlated with brain MAO-A levels.
Of perhaps more immediate clinical interest are studies of one of the serotonin receptors, 5-HT2A. Higher levels of this receptor correlate with lower extracellular serotonin levels. Suicide victims have increased 5HT2A levels in the prefrontal cortex. A variant in the gene for the receptor is strongly associated with response to citalopram, and receptor levels decrease with antidepressant treatment. Further, high levels are associated with pessimism and hopelessness. Meyer references a rating scale, the Dysfunctional Attitude Scale, which he calls the “most promising clinical marker” for 5-HT2A levels.
The serotonin transporter is a protein involved in resorption of serotonin from the synaptic cleft back into the presynaptic neuron, reducing the extracellular serotonin available to the postsynaptic receptor. It is thought to be the site of action of SSRI’s. High serotonin transporter levels in the prefrontal cortex, anterior cingulate, thalamus, caudate, and putamen have been associated with pessimistic, dysfunctional attitudes in patients with major depression. In patients with seasonal affective disorder, high levels are found during the fall and winter.
A different serotonin receptor, 5HT1A, is reduced in affect-modulating regions in individuals with depressive, panic, and social anxiety, though these studies are confounded by the frequent co-occurrence of these disorders. Myer considers the presence of an anxiety disorder a clinical marker of low 5HT1A levels.
Reduction of dopamine release in the putamen is associated with motor retardation, and in the nucleus accumbens, with anhedonia. Most research has studied motor function, since the small size of the nucleus accumbens makes it difficult to image. Meyer identifies motor retardation with reduced dopamine function.
Meyer’s hopes that these finding will make it possible to personalize treatment of depression. A huge problem in clinical psychiatry is that we have array of approved therapies for depression—drugs that address the serotonin, norepinephrine, and/or dopamine neurotransmitter systems; forms of psychotherapy which target cognitive, interpersonal, emotional, and/or experiential problems; electroconvulsive therapy; transcranial magnetic stimulation; vagus nerve stimulation; and experimental approaches such as ketamine, drugs targeting opioid systems, and deep brain stimulation. None of the approved treatments performs consistently better than the others, with the possible exception of ECT. Subtypes like melancholic, anxious, atypical, and even bipolar depression have not been consistently successful at predicting treatment response. Meyer’s association of pessimism, motor retardation, anxiety disorder, and verbal memory deficits with particular neuroimaging abnormalities is interesting and perhaps a small step forward, but these have yet to be associated with response to particular treatments. Many of my own patients have abnormalities in more than one of these clinical domains.
It seems unlikely that any one clinical or biological marker will turn out to be useful at predicting treatment response. What about looking at several and combining them? Back in the 1970’s Joseph Schildkraut and colleagues developed a “D-type Score” which combined several measures of monoamine activity, but its promise of predicting treatment response was never fulfilled. Would adding a PET scan to the clinical evaluation improve treatment selection? Would genotyping make it even better? Perhaps. But the neuroimaging Meyer reviews involved PET scans using different tracers for MAO-A, the dopamine D2 receptor, and the various serotonin-related proteins; magnetic resonance spectroscopy for glutamate; SPECT for the dopamine transporter; and MRI for hippocampal volume. Each of these requires a different isotope and/or scanning machine, and with today’s technology it would be impractical to scan even several of the molecules known to be involved with depression in a clinical population. We can only hope that single scans or other biomarkers will be developed to predict response to particular treatment with sufficient specificity and sensitivity to be clinically useful. I do not have a lot of confidence in this, since several circuits are likely to be involved in even the most narrowly defined subtypes of depression. But a biomarker such as a blood test or a single scan that was sufficiently cheap and nonintrusive might be clinically useful even if it increased our ability to predict treatment response by something like 20%.
My own conclusion from all this is that the clinical heterogeneity of depression is matched by the complexity of its neurobiology, and that we have a long way to go before we have a good understanding how depression works in the brain.
Weinberger DR, Radulescu E, Finding the Elusive Psychiatric “Lesion” With 21st-Century Neuroanatomy: A Note of Caution. Am J Psychiatry 2016; 173:27-33.
Singh MK, Gotlib IH, The Neuroscience of Depression: Implications for Assessment and Intervention. Behav Res Ther 2014; 62:60-73.
Meyer JH, Neuroimaging Markers of Cellular Function in Major Depressive Disorder: Implications for Therapeutics, Personalized Medicine, and Prevention. Clin Pharmacol Ther. 2012 Feb;91(2):201-14