Plasma norepinephrine and prediction of outcome in major depressive disorder

ORIGINAL ARTICLES Plasma Norepinephrine and Prediction of Outcome in Major Depressive Disorder Timothy G. Johnston, Christopher B. Kelly, Michael R. Stevenson, and Stephen J. Cooper Background: Several epidemiologic and clinical factors have been shown to predict long term outcome in major depressive disorder (MDD). The value of biological predictors has not been extensively studied. This study examined whether plasma norepinephrine may be useful in predicting outcome in MDD. Methods: Forty patients were followed up 8 years after an index major depressive episode. Three outcome variables were assessed: time to first recurrence (the primary outcome measure), the Lee and Murray criteria and the Depression Outcome Scale (DOS). The results were examined against plasma norepinephrine value, at the index episode, using survival analysis and linear regression. Results: High plasma norepinephrine at the index episode was positively and significantly associated with time to first recurrence for patients with nonpsychotic MDD (n ⫽ 31, ␹2 ⫽ 8.38, on 1 df, p ⬍ .01). Similarly, plasma norepinephrine was significantly associated with good global outcome, both using Lee and Murray criteria (n ⫽ 34, adjusted R2 ⫽ .24, p ⬍ .01) and DOS criteria (n ⫽ 31, adjusted R2 ⫽ .17, p ⬍ .01) for this group of patients. In contrast, plasma norepinephrine was not significantly related to outcome for MDD with psychotic features. Conclusions: Plasma norepinephrine at index episode seems to be a predictor of outcome in MDD. Biol Psychiatry 1999;46:1253–1258 © 1999 Society of Biological Psychiatry Key Words: Norepinephrine, 3-methoxy-4-hydroxy phenylethyleneglycol, melancholia, depressive disorder, recurrence, outcome

Introduction

M

ajor depressive disorder (MDD) is common in western civilization with a lifetime prevalence of 2.9 –12.6% in studies using DSM III criteria for major depression (Horwath et al 1995). It is a disorder with a From the Department of Mental Health, Whitla Medical Building, The Queen’s University of Belfast, Belfast, UK (TGJ, CBK, SJC); and Department of Medical Statistics, Mulhouse Building, The Queen’s Unversity of Belfast, Belfast, UK (MRS). Address reprint requests to: Dr. Christopher Kelly, Department of Mental Health, The Whitla Medical Building, 97 Lisburn Road, Belfast, UK, BT9 7BL. Received February 1, 1999; revised May 17, 1999; accepted May 21, 1999.

© 1999 Society of Biological Psychiatry

high morbidity and mortality and is the fourth-ranked cause of disability and premature death worldwide (Murray et al 1997). Follow-up studies suggest that the medium to long-term outcome is poor and the mortality risk for suicide is 21 times that of the general population (Harris et al 1998). In a number of follow-up studies several epidemiological and clinical factors have been associated with poor outcome. These include diagnosis of melancholia (Lee et al 1988), older age at onset (Keller et al 1988), three or more previous episodes of depression (Maj et al 1992; Keller et al 1992) the severity of psychopathology (Keller et al 1992) and comorbidity (Mueller et al 1994). Nevertheless a recent prospective study has not been able to confirm these findings and there is inconsistency in the literature (Solomon et al 1997). Though considerable evidence suggests dysfunction of monoamine neurotransmitters in MDD there is not yet any reliable and practical biological marker for prediction of those at greatest risk from ongoing morbidity. Low cerebrospinal fluid (CSF) concentrations of the serotonin metabolite 5-hydroxy-indoleacetic acid (5HIAA) have been associated with suicide and violent attempts at self-harm in MDD (Roy et al 1989). Low CSF 5HIAA is also found to relate to self-harm in other diagnostic groups (Cooper et al 1992). Routine CSF sampling would not be acceptable to most patients. Studies of the norepinephrine (NE) metabolite 3-methoxy-4-hydroxy-phenylethyleneglycol (MHPG) in relation to suicide have found mixed results. One study suggested that low concentrations in both plasma and urine were associated with suicide attempts (Sekunda et al 1992) but this has not been replicated. Studies of CSF MHPG show negative results in relation to suicidal behavior (Sekunda et al 1992; Lester 1995; Mann et al 1997). Because plasma NE correlates well with CSF NE, in both controls and MDD patients (Ziegler et al 1977; Roy et al 1988), this may allow plasma NE to be used as a marker of central activity. There is consistent evidence that plasma NE is raised during an individual episode of major depressive disorder (Kelly and Cooper 1998b; Wyatt et al 1971; Roy et al 1985; Esler et al 1982). Also several studies suggest that electroconvulsive therapy (ECT) reduces plasma NE in MDD patients and that this 0006-3223/99/$20.00 PII S0006-3223(99)00134-1

1254

T.G. Johnston et al

BIOL PSYCHIATRY 1999;46:1253–1258

reduction shows a relationship with short-term improvement of depressive symptoms (Cooper et al 1985; Kelly et al 1997; Werstiuk et al 1996). We hypothesized that a relationship would exist between pretreatment, resting plasma NE and long-term outcome in MDD.

Methods and Materials Forty-seven drug-free depressed patients were recruited between 1988 and 1991, having been diagnosed by two experienced psychiatrists (CBK, SJC) as suffering from major depression. The presence of psychotic and melancholic features were recorded using operational criteria strictly adhering to DSM IIIR (American Psychiatric Association 1987). Severity of depression was rated at index using the Hamilton Depression Rating Scale (HDRS) (Hamilton 1959). Patients were excluded from the study if pregnant, over 70 years of age or suffering from alcoholism, drug addiction, epilepsy, an organic brain syndrome, a neurological disorder, liver disease or serious physical illness. Other illnesses that might alter plasma NE values, such as myxoedema and hypertension were also exclusion criteria. All patients had demographic details recorded including any previous history of depression and a resting sample of blood taken. A standard protocol was adhered to for the collection of all blood samples for plasma NE assay (Kelly et al 1997). Subsequently plasma NE was measured using high-pressure liquid chromatography with electrochemical detection (Kelly et al 1997). Plasma cortisol was measured using radioimmunoassay (Kelly et al 1997). During six months in 1997 and 1998 these patients were followed up by TGJ, who was blind to the biochemical data at index episode. The patients were assessed using a semistructured interview, the SCID I/P (First et al 1996) and from this, the time to first recurrence was recorded. This was defined as the appearance of a new episode of MDD after a six-month asymptomatic period after recovery from the index episode (Frank et al 1991). Also assessed were the Lee and Murray operational criteria for outcome (Lee et al 1988) and the Depression Outcome Scale (DOS) (Surtees et al 1994). The Lee and Murray operational criteria distinguish between four levels of outcome of increasing severity, the levels being pre-assigned to particular combinations of factors. Classification depends, in part, on the frequency that episodes of depressive illness were experienced, whether readmission occurred and the duration of psychotropic medication received. In contrast, the DOS has a list of similar items, but each of these carries equal weight. On both of these scales, the higher the score, the worse the outcome. The study was approved by the local medical ethical committee and each patient gave informed, written consent.

Statistical Analysis Survival analysis, using the Cox Proportional Hazards Model, was used to assess the factors for time to first recurrence (Cox 1972). The key factor was plasma NE. The distribution of plasma NE was skewed to the right and plasma NE was log

transformed to obtain a normal distribution. The following variables were entered into the equation: severity at index; total number of episodes in lifetime; diagnosis at index episode, including comorbid disorders; and the concentration of plasma NE and cortisol at index. Data for MDD patients with psychotic features was analyzed separately from that for the others. This distinction was made based on the balance of current literature suggesting that MDD with psychotic features and MDD alone can be differentiated biochemically, clinically and with regard to treatment response (Johnson et al 1991; Lykouras et al 1994; Schatzberg et al 1992). Within the MDD group the presence of melancholia was used as an additional explanatory variable. Although the Lee and Murray criteria are strictly categorical variables their adoption in this instance as continuous measurements is appropriate for the following reasons. Firstly, they are ordered variables and therefore can lend themselves to be used as a continuous measurement. Secondly, when used in such a manner, they, and the residuals thus formed hold to reasonably good assumptions of normal distributions. Therefore a simple regression equation was created using the Lee and Murray criteria (valued 1– 4), to compare plasma NE level with change in Lee and Murray score. Subsequently the above variables were entered into the equation by using the forward stepwise regression method (using SPSS version 8.0). This procedure was repeated using the DOS criteria as outcome. Separate analyses were performed for the groups of MDD with and without psychotic features

Results Seven of the original 47 patients migrated from the local area during the follow-up period. No interview was possible and no primary care physician notes were available for these 7. Of the remaining 40 patients, 6 were diagnosed as MDD with psychotic features, and were separated out for the purpose of the main analysis, thereby leaving 34 patients as the main study cohort. Demographic, biochemical and outcome variables are shown for the original and main study cohorts in Tables 1 and 2. Of these 34, interviews were only possible with 28, as 4 had died and access for interview was refused in 2 cases (in 1 by the primary care physician and in 1 by the patient). For the six patients who could not be interviewed, there was extensive information available from the hospital and primary care notes and interview was carried out with their psychiatrist and primary care physician. Nevertheless for 3 of these 6 patients (one of whom committed suicide) still insufficient information was available to complete the full DOS. Therefore these 3 were excluded from the analysis of DOS criteria but included in the analysis for Lee and Murray criteria and time to recurrence. These original 34 patients were followed-up for a mean duration of 8 years. During this period, 2 patients died by suicide, 2 from natural causes, 7 had no recurrences and 3

Plasma NE and Prediction of Outcome in Depression

BIOL PSYCHIATRY 1999;46:1253–1258

1255

Table 1. Demographic and Biochemical Baseline Variables for Original and Depressive Cohort Value at index episode Variable Age [mean (SD)] Gender (m/f) Admission status (inpatient/outpatient) Plasma NE [pg/ml, mean (SD)] HDRS [mean (SD)] Plasma cortisol [␮g/ml, mean (SD)]

Original cohort n ⫽ 47

MDD study patients n ⫽ 34

50 (17) 17/30 16/31

47 (8) 10/24 13/21

399 (410)

430 (308)

27 (7)

25 (9)

455 (155)

488 (196)

NE, Norepinephrine; HDRS, Hamilton Depression Rating Scale.

patients did not recover from their index episode. Because these latter 3 patients continued to suffer from their index episode of depression they could not be entered into the survival analysis but could be entered into the regression analyses. Of the remaining 20 patients, 15 had a number of recurrent episodes (range 2–12). Five had their diagnoses revised from MDD: in 4 cases the patient had hypomanic/ manic episodes and were reclassified as suffering from bipolar affective disorder; in one the diagnosis changed to dementia after he developed global cognitive impairment a considerable time after the index episode.

Time to Recurrence For the non-psychotic MDD patients (n ⫽ 31), the plasma NE concentration was positively and significantly associated with time to first recurrence (␹2 ⫽ 8.38, on 1 df, p ⬍ .01): the higher the concentration of plasma NE the longer the time to first recurrence. A scatterplot of this data is shown in Figure 1. No other associations were statistically significant. The hazard function was calculated for this same group using the Cox Proportional Hazards Model (Cox 1972). This indicated that for any 3-fold increase in

Table 2. Outcome Variables for Original and Depressive Cohort Mean/Median (Range) Outcome variable Mean time to recurrence in months Lee and Murray criteria, median score Depression Outcome Scale, median score

Original cohort n ⫽ 47

MDD study patients n ⫽ 34

48 (6 –96)

47.5 (6 –96)

2 (1– 4)

2 (1– 4)

1.5 (0 – 6)

2 (0 – 6)

Figure 1. Scatterplot of log transformation of norepinephrine (LNNE) vs. Time to recurrence for major depressive disorder (n ⫽ 31).

plasma NE the risk of recurrence, in any unit time, is reduced to one-third of its former value.

Global Outcome Using Lee and Murray criteria for the non-psychotic MDD patients the index concentration of plasma NE showed a significant negative association (n ⫽ 34, adjusted R2 ⫽ 0.24, p ⬍ .01) with global outcome. This relationship is shown in Figure 2. Given that higher Lee and Murray scores indicate poorer outcome, this means that the larger the plasma NE, the better the outcome. None of the other variables entered demonstrated statistical significance. Similarly using the DOS for this group of patients, plasma NE demonstrated a significant negative relationship (n ⫽ 31, adjusted R2 ⫽ 0.17, p ⬍ .01) between plasma NE and outcome. Again, the higher the plasma NE the better the outcome. This relationship is shown in Figure 3. No other variables were significantly related. For the MDD group, when melancholia was entered as an explanatory variable no significant association between melancholia and outcome was found. This was evident for time to recurrence and both measures of global outcome. Although the number of patients in the MDD with psychotic features group was small, it was felt necessary to examine if any trends were present in this group. The subsequent analysis, using either time to recurrence or global outcome demonstrated that none of the variables, including plasma NE, reached statistical significance or demonstrated a strong trend.

1256

BIOL PSYCHIATRY 1999;46:1253–1258

Figure 2. Global outcome (Lee and Murray criteria) vs. log transformed NE (Ln NE) for major depressive disorder (n ⫽ 34). Dotted lines represent 95% confidence intervals.

Discussion The results demonstrate a significant relationship between index resting plasma NE and the three outcome measures for patients with MDD, using both survival analysis and regression analysis. Although these measures assess different elements of outcome, namely time to first recurrence and long term outcome, it is of note that plasma NE is related to both. The magnitude of this relationship (i.e., the risk of recurrence falls by a factor of 3 for a 3-fold increase in plasma NE) suggests that this may be clinically significant. This is the first time to our knowledge that such a relationship has been found. This finding is consistent with other studies that have suggested disturbance of the noradrenergic system in MDD (Kelly et al 1997; Kelly et al 1998a). We have previously found that patients in an individual episode of MDD, before treatment, have higher plasma NE than patients with anxiety disorders or control subjects (Kelly et al 1998b). This is consistent with the vast majority of the previous literature (Kelly et al 1998b). In the same study we demonstrated that those in such an episode of MDD also had greater variability in plasma NE whether they were of the melancholic sub-type or not. Depressed patients with the highest plasma NE concentrations may be those with the most “plastic” or mutable noradrenergic systems. We would propose that those with the most “plastic” noradrenergic systems should be those most likely to respond well to treatment and have a better long-term outcome, as the current study demonstrates. ECT treatment results in a fall in plasma NE (Kelly et al

T.G. Johnston et al

1997; Werstiuk et al 1996) that again is consistent with the proposition that those with more mutable noradrenergic systems will respond well to an effective treatment that can influence it. The proposition that plasma NE may provide some measure of treatment response and outcome does not imply that MDD is due primarily to a noradrenergic disturbance. The findings are equally compatible with another biological disturbance being the primary cause with subsequent ability to dysregulate noradrenergic systems being an indicator of one dimension of the illness that includes long term outcome. There are a number of limitations to this study. Firstly there is the difficulty in knowing precisely how elevated plasma NE relates to brain function. There are many intervening steps between activation of the locus coeruleus, its projections and activation of peripheral noradrenergic sympathetic neurones and it would seem dangerously simplistic to argue that the activity of one reflects the activity of the other. Studies have shown plasma NE to correlate well with CSF NE in controls and in MDD patients. More recently, studies in hypertensive patients, examining plasma NE spillover from the brain, show a clear correlation with isolated sympathetic nerve firing rate as measured by microneurography (Esler et al 1995). Although acknowledging the complexity of the relationship, this would suggest that indices of peripheral sympathetic nerve activity could reflect some aspects of central function. An additional limitation is the lack of control for

Figure 3. Global outcome (Depression Outcome Scale criteria) vs. log transformed NE (Ln NE) for major depressive disorder (n ⫽ 31). Dotted lines represent 95% confidence intervals.

Plasma NE and Prediction of Outcome in Depression

antidepressant treatment. All patients, but 1, received antidepressant treatment during their index episode. The treatment duration was for a minimum of four months and the mean Composite Antidepressant Score (CAD; Keller et al 1992) was high (3.1), indicating intense initial treatment. The full range of the CAD is from 0 to 4. A score of 3 is equivalent to a daily dose of 200 mg of imipramine. The information was available, with confidence, on 28 of the MDD patients and when the CAD was analyzed (using survival and regression analyses) there was no statistically significant association with time to recurrence or global outcome. Because of the long period of follow-up we were unable to ascertain the duration and length of prophylaxis with sufficient certainty to allow meaningful analysis of treatment effects on outcome. Whereas it might seem intuitive that increased dose and duration of prophylaxis should be associated with better outcome, naturalistic studies have suggested a more complex relationship between treatment and outcome. This may arise because although some patients do better on high dose, long-term prophylaxis others, who are treatment resistant, with poor outcomes, also receive intensive treatment over a longer period (Brugha et al 1992). Of the original cohort, 41 patients had a MDD without psychotic features. Interviews were possible with 28 and extensive information was available, from medical and psychiatric personnel and notes, to allow an assessment of time to recurrence and global outcome for a further 6. Although the sample of patients is not large, nevertheless it is closely comparable to those in other similar outcome studies. For example outcome measures, such as the rate of chronicity (10%), the probability of remaining well (17%) and the rate of psychotic features (14%) were similar to those found in other longitudinal studies of similar length (Lee et al 1988; Maj et al 1992; Johnson et al 1991). In conclusion, this is the first time a biological marker has been shown to be related to recurrence and morbidity in such a long term follow up of depressive illness. Noradrenergic systems are disrupted in these patients and a high plasma NE seems to be related to longer time to recurrence of illness and lower morbidity, perhaps reflecting a greater ability to respond to treatments.

References American Psychiatric Association (1987): Diagnostic and Statistical Manual of Mental Disorders, 3rd ed rev. (DSM IIIR) Washington, DC: American Psychiatric Press. Brugha TS, Bebbington PE, MacCarthy B, Sturt E, Wykes T (1992): Antidepressants may not assist recovery in practice: a naturalistic prospective survey. Acta Psychiatr Scand 86: 5–11.

BIOL PSYCHIATRY 1999;46:1253–1258

1257

Cooper SJ, Kelly JK, King DJ (1985): Adrenergic receptors in depression. Br J Psychiatry 147:23–29. Cooper SJ, Kelly CB, King DJ (1992): 5-Hydroxyindoleacetic acid in cerebrospinal fluid and prediction of suicidal behaviour in schizophrenia. Lancet 340:940 –941. Cox D (1972): Regression models and life tables. J R Stat Soc 34:187–202. Esler M, Turbott J, Schwarz R, et al (1982): The peripheral kinetics of norepinephrine in depressive illness. Arch Gen Psychiatry 39:295–300. Esler MD, Lambert GW, Ferrier C, et al (1995): Central nervous system noradrenergic control of sympathetic outflow, in normotensive and hypertensive humans. Clin Exp Hypertens 17:409 – 423. First MB, Spitzer RL, Gibbon RL, Williams JB (1996): User’s Guide for the Structured Clinical Interview DSM-IV Disorders-Patient Edition (SCID-I/P, Version 2.0) Biometric Research Department, New York State Psychiatric Institute, 722 West 168th Street, New York. Frank E, Prien RF, Jarrett RB, et al (1991): Conceptualization and rationale for consensus definitions of terms in major depressive disorder. Arch Gen Psychiatry 48:851– 855. Hamilton M (1959): Development of a rating scale for primary depressive illness. Br J Soc Clin Psychiatry 6:278 –296. Harris EC, Barraclough B (1998): Excess mortality of mental disorder. Br J Psychiatry 173:11–53. Horwath E, Weissman MM (1995): Epidemiology of depression and anxiety. In: Tsuang MT, Tohen M, Zahner GP, editors Textbook in Psychiatric Epidemiology. New York: Wiley, 317–344. Johnson J, Horwath E, Weissman MM (1991): The validity of major depression with psychotic features based on a community study. Arch Gen Psychiatry 48:1075–1081. Keller MB, Lavori PW, Lewis CE, Klerman GL (1988): Predictors of relapse in major depressive disorder. JAMA 250: 3299 –3304. Keller MB, Lavori PW, Mueller TI, et al (1992): Time to recovery, chronicity and levels of psychopathology in major depression. Arch Gen Psychiatry 40:809 – 816. Kelly CB, Cooper SJ (1997): Plasma norepinephrine response to electroconvulsive therapy in depressive illness. Br J Psychiatry 171:182–186. Kelly CB, Cooper SJ (1998a): Plasma norepinephrine response to a cold pressor test in subtypes of depressive illness. Psychiatric Res 81:39 –50. Kelly CB, Cooper SJ (1998b): Differences and variability in plasma norepinephrine between depressive and anxiety disorders. J Psychopharmacol 12:161–167. Lee AS, Murray RM (1988): The long-term outcome of Maudsley depressives. Br J Psychiatry 153:741–751. Lester D (1995): The concentration of neurotransmitter metabolites in the cerebrospinal fluid of suicidal individuals: A meta-analysis. Pharmacopsychiatry 28:45–50. Lykouras L, Markianos M, Hatzimanolis J, Mallaras D, Stefanis C (1994): Biogenic amine metabolites in delusional depression and melancholia subtypes of depression. Prog Neuropsychopharmacol Biol Psychiatry 18:1261–1271. Mann JJ, Malone KM (1997): CSF amines and higher-lethality

1258

BIOL PSYCHIATRY 1999;46:1253–1258

suicide attempts in depressed inpatients. Biol Psychiatry 41:162–171. Maj M, Veltro F, Pirozzi R, Labrace S, Magliano L (1992): Pattern of recurrence after recovery from an episode of major depression. Am J Psychiatry 149:795– 800. Mueller TI, Lavori. PW, Keller MB, et al (1994): Prognostic effect of the variable course of alcoholism in the 10-year course of depression. Am J Psychiatry 151:700 –706. Murray CJ, Lopez AD (1997): Global mortality, disability, and the contribution of risk factors: Global Burden of Disease Study. Lancet 349:1436 –1442. Roy A, Pickar D, Linnoila M, Potter WZ (1985): Plasma norepinephrine levels in affective disorder. Arch Gen Psychiatry 42:1181–1185. Roy A, Pickar D, Dejarg J, Karoum F, Linnoila M (1988): Norepinephrine and its metabolites in cerebrospinal fluid, plasma and urine. Arch Gen Psychiatry 45:849 – 857. Roy A, De Jong J, Linnoila M (1989): Cerebrospinal fluid monoamine metabolites and suicidal behaviour in depressed patients. Arch Gen Psychiatry 46:609 – 612. Schatzberg AF, Rothschild AJ (1992): Psychotic major depres-

T.G. Johnston et al

sion: should it be included as a distinct syndrome in DSM IV? Am J Psychiatry 149:733–745. Sekunda SK, Cross C, Koslow S, Katz MM, Kocsis JH, Maas JW (1992): Studies of amine metabolites in depressed patients. Relationship to suicidal behaviour. Ann NY Acad Sci 3:231– 241. Solomon DA, Keller MB, Leon EC, et al (1997): Recovery from major depression: A 10-year prospective follow-up across multiple episodes. Arch Gen Psychiatry 54:1001–1006. Surtees PG, Backley C (1994): Future imperfect; the long-term outcome of depression. Br J Psychiatry 164:327–341. Werstiuk ES, Coote M, Griffith L, Shannon H, Steiner M (1996): Effects of electroconvulsive therapy on peripheral adrenoreceptors, norepinephrine, MHPG and cortisol in depressed patients. Br J Psychiatry 169:758 –765. Wyatt RJ, Portnoy B, Kupfer DJ, Engelman K (1971): Resting plasma catecholamine concentrations in patients with depression and anxiety. Arch Gen Psychiatry 24:65–70. Ziegler MG, Lake CR, Wood JH, Brooks BR, Ebert MH (1977): Relationship between norepinephrine in blood and cerebrospinal fluid in the presence of a blood-cerebrospinal fluid barrier for norepinephrine. J Neurochem 28:677– 679.