Recent Studies on Norepinephrine Systems in Mood Disorders

Alan F. Schatzberg and Joseph J. Schildkraut

That the catecholamine (CA), norepinephrine, may play a pivotal role in the mechanism of action of antidepressant drugs and the pathophysiology of depressive disorders was hypothesized nearly 30 years ago (66). Since that time, research on various aspects of these hypotheses has provided us with much important information about the biology and treatment of depressive disorders. However, many questions remain about the exact role this neurotransmitter plays in depression. In this chapter, we review where we are today and provide a framework for approaching future studies of this important CA system.

Norepinephrine (NE) is synthesized in a variety of peripheral and central sites, including the sympathetic nervous system, adrenal glands, and brain (locus coeruleus). There are limitations in studying this system in man, since most studies of NE or its metabolites in patients and controls involve at best indirect measures of central activity. For example, although metabolite levels in urine or blood may give us important information about central activity, they are still heavily derived from the periphery. Neuroendocrine challenges, such as growth hormone response to clonidine, can also provide potentially important information concerning brain NE activity, but they too are only indirect assessments of NE physiology. The limitations of such approaches has led to debate about the significance of NE in depression, a state of affairs common to research on all other neurotransmitter and neuromodulator systems in psychiatric disorders. The use of postmortem brain tissue can provide useful information, but here too such methodological issues of when and how the tissue was obtained, subjects' diagnoses, drug exposure, and so on, limit what one can truly conclude from such investigations. Ultimately, molecular biology and brain imaging may provide more powerful strategies in this area and could help to clarify the significance of data obtained from current clinical research strategies (see also Central Norepinephrine Neurons and Behavior and Noradrenergic Neural Substrates for Anxiety and Fear: Clinical Associations Based on Preclinical Research).

The study of CA physiology in mood disorders was given a major boost when the metabolite 3-methoxy-4-hydroxyphenylglycol (MHPG) was found to be present in both brain and in the periphery and the proportion of urinary MHPG that was derived from central sources was found to be substantial. There has been a great deal of investigation on 24-hr urinary MHPG levels, because they provide a view of NE function over a full day, and early investigators could more easily and reliably determine MHPG in urine than in blood. Although the exact proportion of urinary MHPG that derives from brain is still in debate, some 20% appears to derive from central NE pools (43).

Early case studies demonstrated that in bipolar patients, urinary MHPG levels were lower during the depressed phase and higher during the manic phase than during periods of euthymia (61). These earliest reports did not attempt to control for the nonspecific effects of activity. Subsequently, we and others reported that urinary MHPG levels were significantly lower in bipolar depressives than in unipolar depressions or control subjects (7, 17, 60, 63). Differences between bipolars and control subjects did not appear to be caused by differences in relative retardation, agitation, or anxiety (65). Generally, these studies included small samples such that the effects of age (and to some extent, sex) were not well explored.

In more recent years, the use of CA measures to distinguish between unipolar and bipolar depressed patients has continued to be a major focus of study. After our group again reported that bipolar depressed patients demonstrate significantly lower urinary MHPG levels than do unipolar nonendogenous subjects (54), several, but not all, recent studies have supported this finding (2, 23, 38, 57). A comparison of these four studies indicates that the nature of the patient's bipolar history (hypomania vs. mania) may be an important variable in determining urinary MHPG levels in bipolar depressed patients. The development of more comprehensive classifications of bipolar disorders, in part on the basis of severity of manic episodes (i.e., bipolar I vs. bipolar II), has aided greatly in this area. In the Depression Collaborative Study, significant differences in catecholamine excretion were not observed when unipolar and bipolar depressives were compared (23). In that study, bipolar patients were not further classified into bipolar I and bipolar II subtypes. In a Swedish study (2), significant differences were also not observed between unipolar and bipolar depressives, but, in that study, over 80% of the bipolar patients appeared to have a bipolar II disorder. In an National Institute of Mental Health (NIMH) intramural study, Muscettola and colleagues (38) reported that patients with bipolar I depressions demonstrated significantly lower urinary MHPG levels than did patients with unipolar depressions. In contrast, significant differences were not observed between bipolar II and unipolar subjects. Similarly, our group (57) reported that bipolar I, but not bipolar II, depressives excreted lower mean 24-hr urinary MHPG levels than did unipolar depressed patients. Thus, the degree of bipolarity may play a crucial role in determining relative urinary MHPG excretion. Unfortunately, the study of bipolar I depressed patients today is difficult since the majority of such patients are on thymoleptic agents and the withdrawal of such agents poses great ethical and medical dilemma.

Plasma NE levels have been reported to be higher in depressives in the supine position than in healthy controls (11, 42, 46). Using orthostatic challenges, unipolar and bipolar patients both demonstrate greater increases in plasma NE after moving to the upright position than do controls (49, 51). In the studies from the NIMH intramural program, heart rate and blood pressure did not change in parallel with plasma NE fashion, but those investigators have suggested that some depressed patients demonstrate insufficient NE tone, that is they require greater changes in NE levels to maintain homeostasis (2, 51).

Plasma NE and MHPG levels also have been compared in unipolar and bipolar patients. Plasma NE levels have been reported to be significantly lower in bipolar patients with a history of melancholia than in their unipolar counterparts who had significantly greater NE levels than control subjects (49). Similar differences between unipolars and bipolars were reported on plasma MHPG measures (47). Degree of bipolarity was not reported in these studies.

Several groups, including our own, have explored the potential application of measures of CAs and other metabolites, in addition to MHPG, to help discriminate among subtypes of depression. In the Depression Collaborative Study, Koslow et al. (23) reported that compared to healthy control subjects, depressed patients demonstrated significant elevations in urinary levels of both NE and epinephrine (E) as well as their metabolites with the exception of MHPG. They suggested that total body CA turnover may provide more salient information about CA dysregulation than does urinary MHPG alone. Increased levels were interpreted as reflecting increased sympathetic nervous system and adrenal activity. In this study, urinary NE levels were significantly higher in unipolar than in bipolar depressed patients.

Maas et al. of the Depression Collaborative Study (26) have reported that the individual ratios of NE to NE plus metabolites or E to E plus metabolites may provide better estimates of CA metabolism than are obtained using individual amine or metabolite data alone. Specifically, they noted that the ratios of NE to NE plus metabolites and E to E plus metabolites were significantly higher in depressed patients than in control subjects. In contrast, the ratio of MHPG to NE plus metabolites was lower in depressives than in controls. They argued that there was a relative increase in CAs and a relative decrease in MHPG in depression. Moreover, they noted that the relative increases in NE and epinephrine were largely due to differences between unipolars and controls. These data not only point to the use of CA measures to help discriminate unipolars from controls but also again suggest such patients may be characterized as having increased sympathetic and adrenal activity.

A number of years ago we reported on the use of discriminant function analysis of 24-hr urinary CAs and metabolites [so-called Depression (D)-type scores] to provide better separation between bipolar manic–depressive from unipolar nonendogenous patients than do urinary MHPG levels alone (62). In a more recent study (57), we reported that bipolar I patients demonstrated both significantly lower MHPG levels and D-type scores than did unipolar nonendogenous patients. However, D-type scores provided greater sensitivity and specificity for separating bipolar I depressions from unipolar nonendogenous depressions (as well as all other subtypes of depressions) than did urinary levels of MHPG, NE, normetanephrine (NMN), vanillylmandelic acid (VMA), or the sum of NE plus its metabolites (see Table 1). In contrast, bipolar II and unipolar depressed patients could not reliably be separated using MHPG or any of the other measures. In this study, age, sex, hospital status, anxiety, and overall severity did not account for differences in D-type scores between bipolar I and unipolar patients. These data suggest that urinary CAs can be used to discriminate among subtypes of depressed patients, particularly those with bipolar I and unipolar subtypes; however, merely summing NE and its metabolites does not provide as powerful discrimination as does using discriminant function analysis of CAs and their metabolites.

The pathophysiological significance of D-type scores remains to be elucidated. It is possible that the equation is correcting for that proportion of urinary MHPG that is not derived from brain. Alternatively, such analysis may be incorporating relevant data provided by measures of urinary NE or its other metabolites. The application of such equations by others will require each laboratory standardizing their assays and determining either specific coefficients or their own equations (42, 44).

In the 1970s, we noted in a preliminary sample that unipolar patients were more heterogeneous with regard to urinary MHPG excretion than were bipolar subjects (63). In the early 1980s, we reported in a larger sample that unipolar patients were relatively heterogeneous in their excretion of urinary MHPG, with some patients demonstrating relatively low values (similar to bipolar patients) others demonstrating very high values, and still others demonstrating midrange values (54). In contrast to bipolar and control comparisons, the mean urinary MHPG level did not differ significantly when unipolar patients and healthy controls were compared. Bipolar and control differences were not due to apparent age or sex effects. At that time, we suggested that unipolar depressives with low or high urinary MHPG levels might represent two different forms of CA dysregulation—the former a low output state and the latter a high output state secondary to either noradrenergic receptor subsensitivity or increased acetylcholine (ACh) activity that could result in increased 24-hr urinary CA and metabolite levels. [More recent studies suggest that elevations in MHPG and cortisol could be due to increased corticotropin-releasing factor (CRF) activity (see below).] The midrange subgroup was viewed as having a dysfunction of another neurotransmitter system. As indicated above, other groups have also reported higher NE and metabolite levels in unipolar depressed patients than in control subjects.

CATECHOLAMINE LEVELS AND RESPONSE TO ANTIDEPRESSANTS

A number of early studies suggested that 24-hr urinary MHPG levels could predict response to antidepressant agents (25, 64). The earliest of these studies reported that patients with low urinary MHPG levels responded to imipramine (25). Subsequently, many studies have reported that low MHPG patients respond significantly more robustly to treatment with tricyclic and tetracyclic agents (imipramine, nortriptyline, and maprotiline) that exert pronounced effects on norepinephrine reuptake than do patients with high urinary MHPG levels (19, 27, 55). In a recent review, Garvey et al. (15) concluded the data were most compelling for MHPG predicting imipramine response. Still, studies have not reported on which drug strategies might prove particularly effective in patients with high MHPG excretion who fail to respond to these agents.

Three nontricyclic antidepressants have been recently studied with respect to relative efficacy in low and high MHPG output depressions. Pretreatment urinary MHPG levels did not predict response to the monoamine oxidase inhibitor, phenelzine, in a group of 38 unipolar patients (67). In another study, our group reported that patients with high urinary MHPG levels responded significantly better to alprazolam, a triazolobenzodiazepine, than did patients with low MHPG levels (36). High, baseline CA excretion in these patients was associated with agonist nonspecific heterologous desensitization of the platelet receptor–G-protein–adenylate cyclase complex (see below). In a recent study, our group has observed that patients with low MHPG levels responded more robustly to treatment with fluoxetine, a selective serotonin reuptake inhibitor, than did patients with high MHPG levels (59).

The three studies above (36, 59, 67) also explored the effects of treatment on urinary NE or MHPG excretion. In all three, significant reduction in CA or metabolite levels were observed in both responders and nonresponders to all agents, and thus reductions did not appear to correlate with treatment response (36, 59, 67). However, in a previous report of the Depression Collaborative Study, amitriptyline and imipramine decreased 24-hr urinary MHPG levels in both unipolar responders and nonresponders, but responders demonstrated greater reductions in MHPG excretion than did nonresponders (5). In our study on alprazolam, the reduction in CA and metabolite excretion in high CA patients was associated with normalization of the desensitized platelet receptor–G-protein–adenylate cyclase complex (36). These data suggest response to treatment with alprazolam may require both a decrease in CA and metabolite output and receptor–G-protein–adenylate cyclase reregulation. Taken together, these studies suggest that reductions in CA and metabolite excretion may be necessary, but not sufficient, for response to antidepressants (see the section below on "Norepinephrine Measures and Mania").

Although in our hands both desipramine and fluoxetine resulted in significant decreases in NE metabolite excretion, longitudinal biochemical data suggest the two agents exert different effects on NE excretion (59). Desipramine was associated with significant reductions in urinary MHPG and VMA by 6 weeks but with significant increases in NE excretion, in keeping with its effects on blocking NE reuptake. In contrast, although fluoxetine resulted in significant decreases in MHPG, VMA, and NMN excretion (albeit for MHPG and VMA less than were seen with desipramine), NE excretion was not increased. These data suggest fluoxetine affects NE excretion through mechanisms that do not involve norepinephrine reuptake blockade. Other studies have reported that selective serotonin reuptake inhibitors reduce levels of NE or metabolites (43), although, not all studies agree (21).

Combining data from our study comparing desipramine (DMI) and fluoxetine with those of other studies, we have noted that the desipramine-induced decrease in urinary MHPG excretion at 4 and 6 weeks was observable after one week of treatment (Schildkraut et al., unpublished data). In contrast, NE and NMN levels demonstrate a more variable response. At 1 week, they are decreased, but, at weeks 4 and 6, they are increased over baseline. These data indicate that response to DMI involves a complex reregulation of NE metabolism over time.

Karege et al. (22) reported that dividing of plasma MHPG levels into three ranges may provide more useful longitudinal information than using group mean values or dividing the sample into high and low subgroups. Low and high plasma MHPG patients who responded to imipramine or desipramine demonstrated "normalization" of their values (i.e., low values increased into normal midrange and high values decreased into the midrange). In contrast, in the overall group, mean MHPG levels were found to decrease with treatment with little relationship to treatment response. These data suggest that changes in MHPG levels may indeed be important for treatment response and that analysis of overall group data may provide less meaningful information than dividing the sample into biochemical subgroups. Further studies on larger samples are required to determine the effectiveness of this strategy.

Mooney et al. (37) have compared urinary MHPG levels and D-type scores to discriminate between responders and nonresponders to alprazolam or imipramine. Patients with low MHPG levels responded better to imipramine than did high MHPG subjects; as indicated above, the converse was true for alprazolam. However, D-type scores provided significantly better discrimination between responders and nonresponders to both drugs than did MHPG levels. These data further suggest that more useful clinical data may be derived from analysis of CAs and their various metabolites than can be obtained using urinary MHPG levels alone (see above).

In the 1970s, Shopsin et al. (68) reported on the potential use of CA depleters to reverse antidepressant responses. This approach has been readapted in recent years by Delgado and colleagues (8). In a series of studies, tryptophan depletion reversed the antidepressant response in patients who had been treated with selective serotonin reuptake inhibitors. Patients who were treated with desipramine were not similarly affected (9).

In a later study, this group administered a-methyl-para-tyrosine (AMPT) to patients treated with desipramine or SSRIs (8). This study involved limited samples; however, data suggest AMPT produces relapses into depression in patients treated with desipramine but not with SSRIs. In contrast to earlier studies in which AMPT may by itself exacerbate depressive symptoms in depressed patients, this group has reported that AMPT does not worsen depressive symptoms (34). Taken together, these data suggest that maintaining NE tone or levels is necessary for sustaining antidepressant responses to potent noradrenergic agents as does maintaining 5-HT levels for sustaining antidepressant responses to SSRIs. They also suggest that these systems play important roles in mediating antidepressant responses.

A number of studies have attempted to correlate clinical characteristics to NE measures. Samson and colleagues (52) recently reported that psychomotor retardation was correlated linearly with MHPG excretion in a group of nonbipolar I depressed patients. In contrast, sleep disturbance [as measured using the Hamilton Depression Rating Scale (HDRS) ratings] was highest in patients with low or high MHPG values. Patients with midrange values demonstrated less sleep disturbance than did low or high MHPG subjects. When data were analyzed dividing the sample into low and high MHPG subgroups, meaningful relationships were not observed between sleep and MHPG excretion. These data further suggest that there may be three biochemical subgroups that can be defined using MHPG levels.

In another study, Samson et al. (53) reported that depressed patients with high MHPG levels were also characterized by perceptions of powerlessness. These findings in depressed patients appear to parallel studies on lower animals relating learned helplessness to depression. The findings of Roy et al. (17) that plasma MHPG correlates significantly and positively with anxiety suggest a possible triad of a sense of powerlessness, anxiety, and elevated MHPG excretion in depressed patients. Further studies of larger samples are required to explore the relationship between symptoms and measures of CA activity.

In the early 1980s, three groups reported on significant positive relationships between measures of CAs and cortisol activity. Stokes et al. (71) reported significant positive correlations between urinary MHPG and plasma cortisol in depressed patients and our group reported significant positive relationships between urinary MHPG and urinary free cortisol (UFC) levels (45). Jimerson et al. (20) reported significant positive relationships between plasma MHPG levels and dexamethasone suppression test (DST) nonsuppression. These results were initially quite surprising since CAs were then thought to act primarily as tonic inhibitors of the hypothalamic–pituitary–adrenal (HPA) axis. Subsequent to these early reports, a number of studies have replicated these initial observations (15, 50, 77), although not all studies agree (24).

Other studies have explored NE, epinephrine, and metabolites other than MHPG in an effort to determine whether CA–HPA axis relationships reflect increased peripheral, sympathetic or adrenal activity. For example, Roy et al. (48) reported significant positive correlations between a number of CA and metabolite measures in several tissues [cerebrospinal fluid (CSF), urine, and blood] and cortisol measures in blood. Similarly, Stokes et al. (72) reported significant, positive correlations between 24-hr levels of epinephrine and cortisol, as measured in urine. Taken together, these studies point to activation of peripheral CA systems in depressed patients with pronounced HPA axis activity. The MHPG–cortisol relationships, however, still suggest a relationship between increased activity of both the HPA axis and central NE systems.

Maes and colleagues (28) reported that the significant, positive correlations between 24-hr MHPG and cortisol levels in urine were lost when the effects of 24-hr urinary volume and creatinine were taken out. Our group has performed similar analyses on a large group of unipolar depressives. In our hands, significant positive correlations between urinary MHPG and urinary free cortisol were still present after any possible effects of urinary volume and creatinine were taken out (58). Similarly, we failed to observe any effects for severity. This area warrants further study, although, data from our studies and those of others using measures in blood and CSF suggest that the MHPG–cortisol relationships are not merely due to nonspecific effects of kidney function.

The biological significance of this CA–HPA axis relationship also remains to be determined. One possibility is that the simultaneous elevation of these measures reflects increased adrenal and sympathetic nervous system activity, perhaps indicative of general stress. Another possibility is that elevated CRF activity centrally could result in simultaneous elevation of both systems (28). Such simultaneous elevation could also reflect increased central cholinergic activity, which could result in increased epinephrine, MHPG, and cortisol levels and which has been hypothesized as playing a role in depression. In addition, b-adrenergic agonists have been reported to increase adrenocorticotropin (ACTH) secretion from the pituitary suggesting adrenergic hyperactivity could lead to or exacerbate HPA axis overactivity (28).

Monoamines are metabolized by monoamine oxidase (MAO). The platelets contain MAO of the B form, which primarily metabolizes dopamine and phenylethylamine. However, we have reported a significant correlation between platelet MAO activity and the ratio of NMN plus MN to VMA, and others have reported that infusion of epinephrine results in increased MAO activity. These data suggest MAO activity could provide information relevant to CA activity.

Several studies have reported on significant positive correlations between platelet MAO activity and cortisol measures. Agren and Oreland (1) initially reported significant positive correlations between platelet MAO activity and UFC in a group of unipolar patients. In that study, MAO activity correlated significantly with insomnia ratings. Subsequently, our group reported that patients with high platelet MAO activity were more likely to fail to suppress when challenged with dexamethasone than were their low MAO counterparts (56). Similar results were noted in both men and women, even though women had higher MAO activity than men. A significant positive correlation was observed between platelet MAO activity and 4 pm post-dexamethasone cortisol levels. Total HDRS scores and ratings of insomnia correlated significantly and positively with both platelet MAO activity and 4 pm post-dexamethasone cortisol levels. These data suggest overall severity and insomnia may be related to elevations in MAO and cortisol measures.

Significant positive relationships between MAO activity and post-dexamethasone cortisol measures have now been replicated by several groups (32, 39). The finding appears to be more robust in unipolar than in bipolar subjects and to not be present in healthy controls or schizophrenic patients (39). The significance of these findings in unipolar subjects is unclear. One possibility is that high platelet MAO activity may be a genetically based risk factor for DST nonsuppression if and when patients become depressed. A second explanation is that elevated, circulating CA levels, that are associated with increased HPA axis activity may result in increased platelet MAO activity (56). Still another is that elevated MAO activity could result in low serotonin levels and resultant supersensitive serotonin receptor activity, which could play a role in increased glucocorticoid activity (32). To date, the underlying pathophysiology that accounts for this curious relationship has not been elucidated.

A number of studies in the past decade have examined CA levels in patients with mania. In one longitudinal study, bipolar patients during manic episodes demonstrated significantly higher plasma NE and epinephrine levels than they did when depressed or when in euthymia (29). Plasma CA levels were also higher during mania than they were in control subjects. In that study, patients were on lithium carbonate during the study raising questions concerning a possible confound, although, these findings are quite reminiscent of much earlier studies (see above).

In the Depression Collaborative Study, CSF levels of MHPG and urinary levels of NE were both significantly higher in manic than in depressed patients or controls (74). Significant correlations were observed between these biochemical measures and severity of mania. Manic patients also demonstrated lower ratios of urinary MHPG to total NE plus metabolites and higher ratios of urinary NE to total NE plus metabolites than did control subjects. Although individual pretreatment CA and metabolite values did not differ between responders and nonresponders to lithium carbonate, responders demonstrated significantly lower ratios of MHPG to total NE plus metabolites and higher ratios of VMA to total NE plus metabolites than did nonresponders. Nonresponder values were similar to controls. Lithium significantly reduced NE turnover in responders and nonresponders, leading the investigators to argue that reduction in NE turnover may be necessary but not sufficient for antimanic effects of lithium carbonate. Reductions of CA and metabolite measures were not merely due to reductions in agitation.

In a later report from the Depression Collaborative Study, Swann et al. (75) noted that environmental sensitivity had a significant effect on urinary NE excretion. Manic patients whose episodes were reported to be "environmentally sensitive" demonstrated elevated NE excretion; in contrast, manic patients with autonomous episodes did not. As indicated above, elevated NE activity in manic patients has also been reported to be a positive predictor of lithium response. These data suggest that the elevated NE activity in mania may be related to the effects of external stressors. In the future it would be of interest to simultaneously explore the effects of stress, perceived sense of powerlessness, and CA metabolite measures in both manic and depressed subjects (see above).

Considerable attention has been paid to a2-noradrenergic receptors in depression since presynaptic a2-receptors control release of NE from central neurons. Several approaches have been used to study a2-receptor activity in patients with depression. Binding by agonists, such as clonidine, has generally been reported to be increased in the platelets of depressed patients (13, 40), although this finding has not been consistent across studies (16). Differences in population demographics, extent of drug washout, assay used, and so on, may help to explain these inconsistencies. Binding studies using yohimbine, an a2-antagonist, have failed to show differences in a2-receptor numbers between patients and controls (42). These data suggest that clonidine as an antagonist may provide more important information regarding a2-receptor activity than is obtained using antagonists. However, recent findings indicate that agents with putative a2-adrenergic receptor activity may bind intensely to nonadrenergic sites on platelets (33), suggesting that clonidine binding data may not merely reflect a2-receptor activity (42).

Another approach has been to explore functional aspects of a2 receptors, for example, by determining epinephrine suppression of adenylate cyclase (AC) activity in platelets or a2-agonist induced platelet aggregation. Several studies have reported that CA inhibition of prostaglandin E1 stimulation of AC is reduced in depressed patients (69), although not all studies agree (14). Mooney et al. (36) have reported a blunting of such suppression, which correlated with increased excretion of both NE and epinephrine in urine. However, elevated NE and epinephrine excretion were also associated with blunting of both prostaglandin and NaF stimulation of AC activity suggesting blunted a2-receptor activity reflected agonist nonspecific or heterologous desensitization of the platelet receptor–G-protein–AC complex. Garcilla-Sevilla et al. (14) have reported that a2-agonist-induced platelet aggregation in depressed patients is increased in depressed patients. The discrepancy between increased responses in platelet aggregation and blunted adenylate cyclase responses to clonidine has been thought to be difficult to reconcile (42). These findings do require further study to understand the underlying physiology.

Hypothalamic a2-receptors control release of growth hormone releasing hormone (GHRH) and subsequently release of growth hormone (GH). Clonidine, stimulates GH release in man. In depressed patients, blunted GH responses to clonidine have been observed in many but not all studies (3, 70). Negative studies have frequently used relatively low doses of clonidine (6). Blunted GH responses to desipramine have been reported in both unipolar and bipolar depressives as well as in patients with mania and panic disorder (4, 61). Of particular interest is the observation that the blunted GH response to clonidine may be a trait marker which persists in depressed patients after recovery (35).

Although the blunted GH response to clonidine has been inferred to reflect postsynaptic a2-receptor down-regulation there is also considerable controversy regarding the mechanisms that underlie these observations (6). Clonidine stimulates both pre and postsynaptic a2 receptors, which produce opposite effects on GH release. Moreover, GH release is also controlled by somatomedins and somatostatin, which have been reported to be altered in some depressed patients. Still, the blunted response to clonidine has been a consistent finding in many studies on depressed patients.

Generally, b receptors are postsynaptic. Early reports by Sulser and colleagues (73) noted that in the rat chronic administration of known antidepressants results in down-regulation of b1 receptors in brain. These observations have been pursued in two ways in man. One approach has been to explore b receptors in the cortex of suicide victims. This approach has yielded conflicting results. Some studies have reported higher b-receptor density in suicide victims than in controls (31), whereas other studies have failed to replicate this finding (10). Psychiatric diagnoses, radioligand used, and previous exposure to antidepressants may be key variables that need to be considered when interpreting studies done to date.

The other approach has been to measure either the number of or functional activity of b receptors in lymphocytes or leukocytes. Although there are numerous reports that there are fewer b receptors in the lymphocytes or leukocytes of depressed patients than in healthy controls, many studies have failed to find similar differences (12, 18). As in a2-receptor studies, there are a number of possible methodological explanations for these disparate results.

Functional activity of beta receptors may be measured by exploring adenylate cyclase responses to specific agonists. Several studies have reported relatively decreased responses of adenylate cyclase activity to b agonists in depressed patients as compared with healthy controls (12, 76). These responses appear to normalize with treatment (30). Potter and colleagues (42) have reported that desensitized b receptors may be inversely correlated to certain measures of NE activity, suggesting that blunted b-receptor activity may reflect or be due to increased circulating NE levels (42). Taken together with the findings of Mooney et al. (36), these studies point to the need to simultaneously measure CA output or turnover and receptor activity to better interpret receptor studies.

Data reviewed suggest that disturbances in NE physiology in depression may occur in conjunction with disruption in homeostasis of other neurotransmitters or peptides. In this section, we briefly review three possible interactions. As indicated above, the simultaneous elevation in measures of CA and HPA axis activity could reflect an increase in ACh activity. Increased ACh activity has long been thought to play a role in the pathophysiology of depression. Early on, Janowsky and colleagues hypothesized that depression represented a relative increase in ACh activity and a decrease in NE activity. In contrast, mania represented a relative increase in NE activity and a decrease in ACh activity. A number of lines of evidence suggest elevated ACh activity is involved in depression; however, data from a number of groups suggest elevated ACh activity is associated with both increased NE and HPA axis activity. For example, administration of physostigmine, a procholinergic agent, increases plasma cortisol, plasma epinephrine and CSF MHPG levels.

Similarly, the simultaneous elevation in NE and cortisol activity could reflect increased CRF activity (28). Several studies have reported that CRF is elevated in the CSF of severely depressed patients, and CRF has been reported to increase NE activity in the locus coeruleus. Thus, the simultaneous increase in NE and HPA axis activity might involve this key neurotransmitter or neuromodulator. However, we recently observed in normal controls that administration of ovine CRH results in a significant increase in plasma homovanillic acid (HVA) but not in plasma MHPG. (Posener, Schildkraut, and Schatzberg, unpublished data). Further studies are required to assess the relationships between CRH and NE activity.

g-Aminobutyric acid (GABA) is a third neurotransmitter that may play a role in NE disturbances in depression. Although this widely expressed neurotransmitter has been thought to exert a tonic inhibitory effect on NE systems, recent data suggest that this is not the case. Although GABA may exert an inhibitory effect on NE in some brain regions, recent data from Petty et al. (41) suggest that GABA may in fact facilitate NE activity. They reported that plasma GABA levels are relatively reduced in depressed patients. The role of GABA in mood disorders and its interactions with NE systems is worthy of further study.

In reviewing the recent literature several findings appear to emerge with reasonable consistency.

1. Bipolar I–but not bipolar II–depressed patients can be separated from unipolar patients

on the basis of urinary MHPG levels as well as other catecholamine and metabolite

measures.

2. At least some unipolar depressed patients are characterized by elevated catecholamine

and metabolite measures.

3. In depressed patients, increased levels of CAs and metabolites as well as elevated

platelet MAO activity are associated with increased levels of cortisol or DST

nonsuppression.

4. Patients with low urinary MHPG levels respond more robustly to imipramine (and

perhaps to fluoxetine) than do patients with high MHPG levels.

5. Depressed patients demonstrate blunted GH responses to clonidine during both periods

of depression and remission.

6. Depressed patients demonstrate decreased responsiveness of b receptors to challenges

with specific agonists.

7. Manic patients demonstrate elevated CA and metabolite levels.

8. Treatment of depressed and manic patients, with antidepressants and lithium carbonate,

respectively, results in a decrease in NE turnover.

The significance and explanation of these findings remains to be determined. Although they point to alterations in CA activity in affectively ill patients, it is difficult to determine the relative peripheral and central contributions of these findings, the exact roles that receptor activity and CA synthesis play, and the specific biological dysfunctions that may account for these findings. Alternative approaches need to be developed further to move along research in this area; imaging studies may help to define the key loci of CA activity both for underlying pathophysiology and clinical biological characteristics. Furthermore, the development of animal models with such biological hallmarks as increased NE and HPA axis activity (particularly in response to stress and behavioral sequelae) would enable one to assess the primacy of one or another system in the pathogenesis of depression on their relative contributions. Molecular biological studies of CA regulation are needed to define normative and maladaptive processes and their possible genetic underpinnings. Such studies can be done in both lower animals and in man.

This work was supported in part by grants MH15413 and MH38671 from the National Institute of Mental Health as well as by grants from the Poitras Charitable Foundation and the Karen Tucker Fund.

We also would like to thank Marsha D. Wallace for her assistance in the preparation of this manuscript. A. F. Schatzberg: Department of Psychiatry and Behavioral Sciences, Stanford University School of Medicine, Stanford, California 94305.

published 2000