Neurobiology has informed the genetic search for candidate genes, starting with the human serotonin transporter (SERT) gene (see Chapter 4.5.5). There has been a terrific proliferation of possible genetic effects deriving from neurobiological theories designed either to explain elements of the actions of psychotropic drugs, the depressed phenotype or from animal experiments. The latter are limited by the validity of animal models of depression per se. Some of the former will be noticed below.

Genes making small contributions to the risk of psychiatric disorder are emerging from direct analysis of the genome (see Chapter 2.4.2). Consistent findings must inform biological investigations in future. At this point it is uncertain whether insights will come from studying variation in individual genes, as has often been assumed, or from a much more complex understanding of cellular function regulated only in part by genetic variation. On the latter assumption the role of genetic hits is to direct attention to processes which may go wrong in the relevant disease. For mood disorder, these seem likely to be developmental or related to stress regulation.

The way in which genes may regulate the expression of vulnerability traits is suggested by animal studies. For example, when animals are selected for differences in emotional behaviour they also show different hypothalamic–pituitary–adrenal (HPA) axis function. Specifically, Roman high- and low-avoidance rats differentially acquire a two-way active avoidance response in a shuttle box. High-avoidance animals show greater prolactin and HPA axis responsivity to stress compared with low-avoidance animals. However, young Roman strain rats show identical HPA axis reactivity, although prolactin responses and behaviour are different.⁽¹⁾ In other words, reactivity to the environment may share a measure of common genetic control across physiological and behavioural domains, but HPA abnormality per se develops secondary to emotional experience, or at least is magnified by it.

In human studies, neuroticism is an old psychological construct often criticized as reflecting an average or habitual mood state rather than a truly independent risk. We have studied extremes of the dimension (high and low N) in young subjects before the onset of depression and in older groups who may or may not have experienced depressive episodes. Interestingly, high neuroticism with or without a history of depression is associated with increased awakening cortisol⁽²⁾ in mature subjects, but not in subjects under 20 years of age, echoing the rodent finding. Thus, N has a purely biological consequence that develops with emotional experience, but is independent of depression per se.

What the neuroticism construct has also lacked hitherto has been a plausible psychological dimension. Cognitive bias relevant to the onset of depression can be detected in young high N subjects. In emotional categorization and memory tasks, high N volunteers were faster to classify dislikeable self-referent personality characteristics and produced fewer positive memory intrusions. They also had a higher threshold for identifying happy faces. This suggests the hypothesis that risk for depression is largely manifest as reduced positive processing of emotional information⁽³⁾; increased negative processing appears to develop only after the actual experience of depression. Neural biases underlying this behaviour are even more readily detected.⁽⁴⁾ Our hypothesis is that high neuroticism is not just an habitual low mood but is biologically founded in negative biases in attention, processing, and memory for emotional material. Indeed, there is now genetic evidence favouring a common genetic locus in human beings and rodent.⁽⁵⁾ How emotional bias translates into either low-level symptoms or a full mood episode will be of great interest. Furthermore, depressive episodes per se appear to have an impact on brain function, and increase the risk of further relapse (see below).

Adverse childhood experience was identified in genetically uncontrolled studies as a risk factor predisposing women to subsequent
depression (Chapter 4.5.5) and has been confirmed in genetically informative designs.⁽⁶⁾ In a clinical context, such developmental or social effects are usually viewed as separable from biology. Indeed, their very existence is usually taken to validate a ‘social’ approach to psychiatry. From a more unified point of view, however, one would predict measurable neurobiological consequences. In fact, such effects have proved to be more profound than most biologists anticipated.

Variations in maternal care produce individual differences in neuroendocrine responses to stress in rats. The offspring of mothers that exhibited more licking and grooming of pups during the first 10 days after birth showed, in adult life, reduced plasma ACTH and corticosterone responses to acute stress.⁽⁷⁾ In addition, there was increased hippocampal glucocorticoid-receptor messenger RNA (mRNA) expression, enhanced glucocorticoid feedback sensitivity, and decreased levels of hypothalamic corticotrophinreleasing hormone (CRH) mRNA. Greater early maternal attention also substantially reduced subsequent behavioural fearfulness in response to novelty, increased benzodiazepine receptor density in the amygdala and locus coeruleus, increased α2-adrenoreceptor density in the locus coeruleus, and decreased CRH receptor density in the locus coeruleus. Thus, maternal care serves to programme behavioural responses to stress in the offspring by altering the development of the neural systems that mediate fearfulness.

When BALB/cByJ mice were raised by an attentive C57BL/6ByJ dam, their excessive stress-elicited HPA activity was reduced, as were their behavioural impairments. However, cross-fostering the more resilient C57BL/6ByJ mice to an inattentive BALB/cByJ dam failed to elicit behavioural disturbances. In other words, vulnerable offspring may have their problems exacerbated by maternal behaviour, while early-life manipulations may have less obvious effects in relatively hardy animals.⁽⁸⁾ Whether separation or stress paradigms in rodents can be taken as precise models of the mechanisms underlying the risk of mood disorder or other psychiatric problems cannot yet be decided, but their general relevance to the human case seems obvious. At present, data in human subjects is limited but findings that relate to the better characterized animal models are emerging.⁽⁹⁾

In fact, epidemiological data have linked increased risks of cardiovascular, metabolic, neuroendocrine, and psychiatric disorders in adulthood with an adverse foetal environment as well. Glucocorticoid excess may be the mechanism.⁽¹⁰⁾ Low-birthweight babies have higher plasma cortisol levels throughout adult life, which suggests a permanent change in HPA function. Whether such effects and later effects of environmental stress in childhood can in part mediate co-morbidity between a range of psychiatric and physical disorders is of growing contemporary interest. It is unclear how, over- or underactivity in stress regulation contributes to psychiatric disorder: both appear to be implicated since awakening cortisol responses may be blunted in subjects with early adversity⁽⁹⁾ or enhanced in at risk neurotic individuals.

Gene–environment interaction is the likely basis of the neurobiology of mood disorder. In general terms this must be correct. Either the genetic/biological or the environmental factors could be targets for prevention. Whether the genetic mechanisms can be brought into sufficient focus to allow specific new pathways to be identified remains the major current challenge. It is often assumed that mediating characteristics or the endophenotype may have a simpler genetic architecture than the disease itself: unfortunately, the evidence so far gives reason for caution. This debate is currently very polarized between optimists (see Chapter 2.5.3 by Meyer-Lindenberg & Goldberg) and pessimists (see Chapter 2.4.2 by Flint). The genetic and developmental routes into distal common pathways regulating stress responses may be very numerous. Disorders that are both common and very variable in expression, such as depression, may turn out to have little specificity that is worth talking about. Every illness may be an ensemble of many specific factors, none of which is individually going to lead to a more focused treatment or a better prediction of treatment response.

Perfusion or metabolic imaging can indirectly detect changes in neuronal activity (see Chapters 2.3.6 and 2.3.8). Signals can be well localized, but their meaning is ambiguous. They may reflect either reversible changes in function or a semi-permanent loss of neuronal connectivity. Reductions in function in anterior brain structures have been typical in major depression. Hypoperfusion tends to be greatest in frontal, temporal, and parietal areas and most extensive in older patients; high Hamilton scores tend to be associated with reduced perfusion.⁽¹⁴⁾ Reductions in frontal areas may be more likely in patients with impoverished mental states. Thus, neuropsychological testing in major depression shows evidence of slowing in motor and cognitive domains, with additional prominent effects on mnemonic function that are most marked in the elderly. These effects are correlated with reduced frontal perfusion in the elderly. In younger patients, there may actually be increased perfusion in the frontal and cingulate cortex. Metabolic increases in the cingulate gyrus have been associated with a good treatment response.⁽¹⁵⁾ Highly localizing findings have been unusual, however. The only exceptions have been within-subject changes on recovery in the mesial frontal cortex and perhaps the basal ganglia.⁽¹⁴⁾

There has been a dramatic expansion of imaging studies of emotional processing in normal volunteers, now usually with fMRI (see Chapters 2.3.8 and 2.5.4). It is well summarized by meta-analysis of over 300 such emotion induction and cognitive task. Emotion induction resulted in inferior medial activation and cognitive tasks resulted in dorsolateral activation.⁽¹⁶⁾ However, the broad spread of precise loci of activation means that localization within the frontal lobes has proceeded little further. It may explain the diffuse reports typical of the depression literature. Nevertheless, a focus on limbic activity has led to quite specific, quasineurological hypotheses about connectivity in frontal areas and to treatment innovation: deep brain stimulation adjacent to subgenual cingulated cortex (Brodmann area 25).⁽¹⁷⁾ How effective, and how localized this treatment effect really is, will be an important challenge to the field. However, it underlines that ‘functional imaging’ of brain perfusion primarily informs anatomy.

Isotope-based imaging of receptor occupation could more plausibly offer mechanistic understanding of psychiatric disorder. In depression, it has progressed with the availability of suitably informative ligands. However, the field generally tends to employ small sample sizes, and fundamental advances are difficult to identify. Single-photon emission tomography (SPET or SPECT) with the dopamine D_2/3 ligand [123I]IBZM showed increased binding in the striatum.⁽¹⁸⁾ There were significant correlations between IBZM binding in the left and right striatum and measures of reaction time and verbal fluency, but not of mood as such. This finding has been confirmed with a PET ligand.⁽¹⁹⁾ Increased D_2/3 binding in the striatum probably reflects a reduced dopamine function, whether due to a reduced release or secondary upregulation of receptors. Binding to the 5-HT1a receptor appears to be reduced in unipolar depression, an effect also present in recovered atients.⁽²⁰⁾