Recent Studies of the Biology and Treatment of Depression

Depression is a common disorder. The National Comorbidity Survey reported a prevalence rate of 5% for current depression and a lifetime rate of 17% (1). Some investigators have noted that these rates are higher than those seen in the Epidemiologic Catchment Area (ECA) survey (2), suggesting an overinclusion of milder forms of depression. Indeed, a reassessment and follow-up study using more clinically relevant definitions of severity of illness yielded rates of depression subgroups that are more consonant with other reports as well as clinical impressions (2, 3). Still, major depression is a common disorder not only in the United States but in all societies, and even milder forms carry considerable morbidity (3). Indeed, the World Health Organization/World Bank Study ranked unipolar depression the fourth highest cause of morbidity in 1990, with the expectation that by 2020 it would become the second leading cause of disability (4, 5).

Several recent studies have explored specific subtypes of depression. For example, atypical features as defined in DSM-IV-TR have been a focus of a number of epidemiological and clinical studies (6, 7). Parker and colleagues (6), in Australia, reported that the absence of mood reactivity appeared to be associated with severity of depression and that other atypical symptoms (e.g., hyperphagia) did not appear to be related to each other, raising questions about the validity of the subtype. Others, however, have observed that atypical features do coalesce (7, 8), so this issue remains subject to debate.

Psychotic major depression was the focus of one analysis of a large European survey involving some 19,000 subjects in five countries (9). Psychotic features were found in nearly 19% of subjects who met criteria for a major depressive episode (MDE), a rate slightly higher than the 15% reported in the ECA study. Psychotic features were most commonly seen in individuals who endorsed eight or nine items of the criteria for major depression (33%), but they were also seen in individuals with milder forms of MDE.

Alternative presentations of MDE have increasingly become a research focus, for several reasons, including the common presentation of MDE as physical complaints in primary care and the efficacy of several psychotropic agents in both mood and anxiety disorders as well as chronic pain. Many patients with MDE in primary care settings present with complaints of physical symptoms, including pain (10). Although pain is commonly observed in MDE in primary care settings, there has been considerable skepticism about the generalizability to epidemiological community samples or to psychiatric practices.

In the large European sample mentioned above, chronic painful physical symptoms were observed in 16% of the general population but in 43% of subjects who met criteria for MDE (11). Less than half of the respondents who reported MDE and chronic pain symptoms had an identifiable organic cause of their pain. Headache (including neck pain), shoulder pain, and backache were the most common types of pain. Compared with DSM-IV-TR criteria for MDE, chronic painful symptoms were more commonly seen in subjects with MDE than was guilt, and nearly as commonly as loss of energy. These data have been recently replicated in a study in California (unpublished 2004 study of M. M. Ohayon). Consideration should be given to including chronic pain symptoms in the criteria for major depression in future classification schemes.

As more effective medication and psychotherapy strategies have been developed, greater attention has been given to the significance of residual symptoms that do not meet full criteria for MDE; patients with such residual symptoms have been termed partial responders and their illnesses subsyndromal disorders. Analyzing outcome data from the National Institute of Mental Health (NIMH) Collaborative Program on the Psychobiology of Depression (Collaborative Depression Study), Judd and colleagues (12, 13) reported that residual depressive symptoms are associated with a higher risk of relapse or recurrence, greater utilization of medical services, and greater risk of substance abuse. Thus, increasing attention has been given to maximizing response to both somatic and psychosocial interventions (discussed further below).

A number of explanations have been proposed for the continuation of depressive symptoms despite treatment. Paykel’s group (14) reported a decade ago that residual physical symptoms of depression were associated with a greater likelihood of relapse or recurrence. More recently, Fava and associates (15), in a pooled analysis of studies of duloxetine (a dual norepinephrine/serotonin reuptake inhibitor), reported that remission of depressive symptoms was highly associated with improvement in pain symptoms. Thus, focusing on both physical and emotional symptoms in depression could provide added benefit. Prospective large-scale studies comparing dual uptake agents and selective serotonin reuptake inhibitors (SSRIs) in depressed patients with comorbid pain could help illuminate this area.

Another area that has attracted attention is the role childhood abuse may have in adult major depression and response to treatment. Our group has reported that in chronically depressed patients with early abuse, a specialized form of psychotherapy (cognitive behavioral analysis system of psychotherapy—CBASP) was more effective than nefazodone (16), whereas the opposite was seen in nonabused patients. Early abuse has also been reported to be associated with an increased risk of depression, comorbid pain, and utilization of medical services in health maintenance organization (HMO) settings (17). Thus, research data appear to be converging on an important adult medical phenomenon with roots in childhood.

The genetics of major depression has been a focus of a great deal of research over the past several decades. For much of that time, linkage studies failed to establish clearly any single candidate gene, which led investigators to argue that depression represented a complex genetic disorder, to which many different genes could potentially contribute, either alone or in combination with other genes or with environmental factors such as stress. Recently a number of interesting candidate genes have been identified, primarily from population-based or case-control studies. Several of these genes point to potential interactions with stress.

The short form of the promoter for the serotonin (5-HT) transporter was reported a number of years ago to be associated with neurotic traits. Although subsequent studies did not find a genetic risk of depression in subjects with a short promoter (18), the short form has been associated with increased amygdala activation with stress (19). In the past few years, the short form has also been reported to predict poor response or intolerance to SSRIs in Caucasians in Europe and in the United States (see Pharmacogenetics below). In a major longitudinal study in New Zealand that followed subjects from childhood (18), the short form of the promoter by itself was again not found to be associated with increased risk of depression in the absence of stressors; similarly, stressors by themselves did not predict major depression. However, a significant gene-by-stress interaction was noted, with s/s homozygotes for the transporter promoter at greater risk of developing depression if three or more stressful life events were encountered. This could be a major lead in our understanding of the interaction between biological risk and psychosocial precipitants.

Stress activates both the hypothalamic-pituitary-adrenal (HPA) axis and the sympathetic nervous system. The ability to turn off the HPA axis via feedback mechanisms is widely believed to be a key feature of healthy adaptation, and individuals who are less able to moderate their stress response may be more prone to depression. Feedback inhibition is mediated in part via the glucocorticoid receptor in the hypothalamus, hippocampus, and pituitary. This receptor is surrounded by chaperone heat shock proteins (HSPs), one of which (HSP90) has been implicated in increasing the risk of depressive episode as well as predicting more rapid response to antidepressant treatment. Multiple single-nucleotide-polymorphism genotypes of the FKBP5 gene were explored in two German cohorts. Patients with the T/T genotype at rs1360780 demonstrated more than twice as many depressive episodes as did the C/T or C/C subtypes (20). This T/T genotype is associated with increased expression of the FKBP5 protein, which may result in glucocorticoid receptor insensitivity. However, direct study of patients with the T/T genotype did not point to their having more elevated cortisol levels, and they demonstrated more rapid responses to antidepressant treatment. Thus, although the exact role of this gene or protein in depression remains to be elucidated, the gene is clearly worthy of further study.

Decreased serotonin activity has long been thought to play a key role in the pathophysiology of depression and response to treatment. Serotonin is synthesized from tryptophan via a tryptophan hydroxylase. Recently Zhang et al. (21) and Patel et al. (22) elegantly demonstrated that neuronal synthesis of serotonin is controlled largely via tryptophan hydroxylase 2 (TPH-2) and not the 1 form, which was formerly thought to regulate synthesis in the brain but is now seen to be involved mostly in the periphery. This observation was recently supported by a report that a mutation in TPH-2 was found in some 10% of subjects with unipolar major depression, compared with 1.5% of control subjects (23). This mutation was associated with an 80% decrease in serotonin levels when expressed in a cell line. The variant was not observed in a cohort with bipolar illness. Thus, at least some depressions may be due to a gene variant associated with low serotonin activity. Whether this TPH-2 gene can be used to predict response to SSRI treatment has not yet been studied.

In two of the three genes discussed above, stress could play an important role in interacting with a genetic risk to lead to a depression. In the third gene, decreased synthesis of an important brain neurotransmitter could reflect a specific genetic variant.

Just a few years ago my colleagues and I commented in the fourth edition of the Manual of Clinical Psychopharmacology that the horizon seemed bright for antidepressant agents with new mechanisms of action (57). However, a number of then-promising agents have not received approval from the Food and Drug Administration (FDA), and the near-term horizon seems quite a bit dimmer. Promising approaches are summarized in Table 2.

Pharmacogenetics is becoming a major focus in psychopharmacologic research. This approach uses genetic alterations—often single-nucleotide polymorphisms (SNPs)—to assess the likelihood of response to a particular drug or class of drugs or to predict side effects. The approach has been used in various psychiatric disorders, with perhaps the strongest findings seen in depression. Findings are summarized in Table 3.

Several years ago the Milan group reported that a short form of the serotonin transporter was associated with poor responses to specific paroxetine (89, 90). In contrast, the alternate long form predicted positive responses. This observation has been confirmed in other studies with Caucasians (91), but the opposite prediction pattern has been reported in studies with Asian populations (92). Recently our group reported that the long form of the transporter was only a mild predictor of positive response to paroxetine in geriatric patients (93). In contrast, the short form of the transporter predicted intolerance to the SSRI paroxetine but not to the 5-HT₂ antagonist mirtazapine. Patients who were s/s tolerated mirtazapine much better than did l/l patients (93). Our view of previous studies has been that using last-observation-carried-forward (LOCF) approaches to analysis of patients who dropped out may have resulted in a confusion of intolerance and nonresponse in s/s patients. At any rate, the data do suggest that in Caucasians the SSRI response may not be optimal.

The 102 T/C SNP of the 5-HT_2A receptor has been associated with the response pattern to clozapine (94) in patients with schizophrenia. We explored whether the C/C homozygote for the receptor—seen in some 25% of the population—was associated with response to either paroxetine or mirtazapine in geriatric depressives (95). The C/C homozygocity predicted intolerance to paroxetine, with over 40% dropping out because of adverse events by 8 weeks. In contrast, the C/C form did not predict intolerance to mirtazapine. Response or remission to either drug was not predicted by the 5-HT_2A variants. The s/s form of the transporter and the C/C form of the 5-HT_2A receptor independently predicted intolerance to paroxetine (93). Thus, these data suggest that alterations in 5-HT reuptake or 5-HT_2A postsynaptic receptor activity affect the ability to tolerate an SSRI but do not adversely affect tolerability of a postsynaptic receptor antagonist.

Similarly, allelic variation of the norepinephrine transporter has been explored as a predictor of response to the selective norepinephrine reuptake inhibitor milnacipran in Japanese depressives (96). The T allele of the T-182C polymorphism of the norepinephrine transporter predicted positive response to the drug; the A/A form did not. Allelic variation for the serotonin transporter did not predict response to the norepinephrine agent.

Apolipoprotein E ε4 (APOE-4) has been reported to be a risk factor for Alzheimer’s disease. Individuals with this allele may be at increased risk of greater morbidity after surgery or after head injury and have an increased risk of obstructive sleep apnea. In our geriatric study (97, 98), we explored the hypothesis that APOE-4 alleles were predictors of poor overall response to antidepressant therapy in geriatric patients. This prediction was not borne out, but patients with APOE-4 al-leles were dramatic responders to mirtazapine but not to paroxetine (98). The explanation is not entirely clear, but APOE status may be associated with noradrenergic dysfunction (99) or excessive cortisol activity (100), both of which are targets for mirtazapine therapy (100, 101).

Glucocorticoid receptors are nuclear and are surrounded by chaperone heat-shock proteins. As discussed above, alterations in one HSP have been reported to be associated with multiple depressive episodes and rapid response to drug therapy (20). These alterations may affect the individual’s ability to halt the stress response. Mirtazapine was frequently used in this study, and the prediction of rapid response may have to do with the drug’s ability to lower cortisol levels (101). Further studies in other subject populations will help elucidate this issue.

In regard to pharmacokinetics, two major systems have been the focus of much recent research: P450 2D6 in the liver and medication-resistant P-glycoprotein (mr-P-GP), which controls efflux of drug from the brain. P450 represents a basket of enzymes found primarily in the liver that metabolize various drugs. The best known is P450 2D6. Many drugs serve as substrates as well as inhibitors of P450 2D6, whereas some stimulate activity. There are numerous alleles for 2D6; some connote increased clearance or metabolism, whereas others connote slow metabolism. Slow metabolizers should be more likely to experience side effects. Rapid or ultrafast metabolizers may clear the drug so quickly that they fail to achieve adequate blood levels and are thus less likely to attain a drug effect.

We explored whether slow metabolizers in a geriatric population were more likely to drop out because of side effects of paroxetine (a substrate and inhibitor of 2D6) or of mirtazapine (a substrate but not an inhibitor) (93, 97). We did not observe an effect of 2D6 alleles on dropouts due to adverse events from either drug, although the number of very slow metabolizers (i.e., null alleles) was small. As indicated above, we did observe pharmacodynamic predictors of SSRI response in this study (97).

In the past few years greater attention has been given to mr-P-GP as a predictor of antidepressant response. In mice, knocking out the gene for the pump protein points to its role in facilitating the efflux of both cortisol and antidepressants from the brain (102–105). In studies to date using knockouts, citalopram, amitriptyline, and trimipramine appear to be transported out of mouse brain by P-GP (103–104). Alterations in this gene may tell us more about treatment resistance than does drug clearance via the liver. However, the significance of these observations on P-GP in lower animals to the treatment of depressed patients is unclear, since the knockout model may not be fully relevant to the clinic. Still, this is an interesting area that is fast becoming a focus of pharmacokinetic research.

Recent studies point to key brain areas as having important roles in depression as well as to specific genes as risk factors for developing depression, often in interaction with environmental stress. New antidepressant strategies are being pursued, and despite some recent disappointments in this area, pharmacogenetics is pointing to potential tools for selecting drugs or deciding how to dose them.

Acknowledgment This paper was supported by grant MH 50604 from the National Institute of Mental Health.

CME Disclosure Statement Alan F., Schatzberg, Chairman, M.D., Department of Psychiatry and Behavioral Sciences, Stanford University School of Medicine.

Consultant/Speaker: Abbott, Aventis, Bristol-Myers Squibb, Corcept, Lilly, Forest, GlaxoSmithKline, Innapharma, Janssen, Merck, Novartis, Organon, Pharmacia, Solvay, Somerset, Wyeth. Grants: Bristol-Myers Squibb, Lilly, Wyeth. Equity: Corcept, Cypress Biosciences, Elan, Merck, Pfizer. Co-founder: Corcept Therapeutics. Co-inventor: intellectual property owned by Stanford University.