Portal de Periódicos da CAPES

Should the dexamethasone suppression test be resurrected?

2005; Wiley; Volume: 112; Issue: 4 Linguagem: Inglês

10.1111/j.1600-0447.2005.00628.x

1600-0447

Max Fink,

Tryptophan and brain disorders

Following reports that plasma cortisol levels were elevated in patients with Cushing's disease, similar abnormalities were found in patients with severe mood disorders (1-3). In normal persons adrenal cortisol exhibits a diurnal rhythm, which peaks in the morning and troughs in the early night. The interdependence of the body's hormonal systems encompasses the release of hypothalamic CRH, stimulation of pituitary ACTH, that then evokes adrenal cortisol in a diurnal cycle. The release is inhibited by synthetic glucocorticoids such as dexamethasone and prednisolone. In severely depressed patients plasma cortisol levels are elevated, diurnal rhythmicity is shallow, and the response to synthetic glucocorticoids is lost. Interest in cortisol functioning in psychiatric patients led to the development of the dexamethasone suppression test (DST) as a formal test of hypothalamic-pituitary-adrenal (HPA) function. Interest then waned and little use is made of the DST in clinical psychopharmacology today. In this number of the journal, Yuuki et al. (4) remind us that the abnormal HPA functioning in patients with major depression resolves with adequate treatment. They applied the dexamethasone-corticotropin releasing hormone test (DEX/CRH), a modification of the DST, in their assessment. Much credit for the development of our understanding of the HPA axis and the DST is due to the work of Bernard Carroll and his associates at Universities of Melbourne and Michigan that began in 1967 (2, 3). They studied cortisol release in plasma, urine, and cerebrospinal fluid, varied the time and dosage of glucocorticoid administration, examined different sampling schedules, and optimized the statistical handling of these complex measures (5). They explored the differences between severely depressed and non-depressed psychiatric patients and normal subjects. The result was an illuminating moment in the pathophysiology of mood disorders. Abnormal cortisol regulation and failure to suppress with dexamethasone were frequent in the more severely depressed patients, especially in those with weight loss, agitation, suicide risk, and psychosis. Elevated plasma cortisol levels fell with successful antidepressant treatment, exhibiting posttreatment values similar to those of normal subjects. An example is reported in five depressed patients treated with ECT who exhibited abnormal cortisol functioning that normalized when substantial improvement in clinical condition occurred (5). Of depressed patients studied over four years, seven patients persisted in an abnormal response to dexamethasone despite treatment. All exhibited persistent signs of depressive illness and active out-patient treatments were required to prevent relapse and re-hospitalization. Carroll concluded that: ‘‘The suppression test distinguished clearly between the depressed and the control patients who could not be distinguished simply on the basis of their diurnal plasma cortisol levels’’ (5, p. 104). By 1976, the details of the DST as a measure of severe depression were established (6). In 42 patients with ‘endogenous depression’ compared with 42 patients with other psychiatric disorders, the patients with endogenous depression had greater HPA activity before dexamethasone and less complete HPA suppression after dexamethasone. Patients with two or more abnormal cortisol values after dexamethasone were correctly identified as being endogenously depressed. The authors summarized their findings: ‘‘By using an observation period of 24 h postadministration of dexamethasone, a graded series of abnormal test responses was identified. Depressed patients show abnormal early escape from suppression rather than absolute resistance to HPA suppression by dexamethasone. … The essential disturbance of neuroendocrine regulation in depression is a failure of the normal brain inhibitory influence on the HPA system’’ (6). In 1981, a study validated the DST in 438 patients using a cut-off level of 5 μg/dl plasma cortisol after 1 mg dexamethasone administration ‘‘… as a test for the diagnosis of melancholia (endogenous depression)’’ (7). Two blood samples taken at 16.00 and 23.00 hours after dexamethasone administered at 23.00 hours the night before ‘‘detected 98% of the abnormal test results. This version of the DST identified melancholic patients with a sensitivity of 67% and a specificity of 96%.’’ While the formal test is based on multiple samples, assuring a more stable test, single samples, usually taken at 16.00 hours (4 pm), were also used in testing out-patients (8). Doing the test properly is complex. It requires that the laboratory establish its own normative standards. Plasma dexamethasone levels are often necessary as the test is sensitive to steroid levels. Pregnancy, systemic disease inducing weight loss or requiring endocrine treatment, and starvation are conditions that may affect the test result and must be considered in assessing the test data. After accounting for the variations in test administration and chemical assays, the over-riding conclusion was that abnormal cortisol functioning is a feature of severely depressed patients that normalizes with successful antidepressant treatment (9). The more the patients met criteria for endogenous depressive illness, the higher the percentage of abnormality. When depressed patients are divided into melancholic and non-melancholic groups using a classic formulation for melancholia, high serum cortisol levels and non-suppression to DST is mainly found in the melancholic group. An interesting example is the analysis of the DST in samples divided using the concept of ‘endogenomorphic depression’ (3, 10). Baseline measures of plasma cortisol and urinary free cortisol (UFC) were higher in 40% of 42 patients with endogenomorphic depression compared with elevated levels in 12% of the 42 patients with depressive neuroses. The probability of a patient with a high value being an endogenomorphic depressive was calculated to be 77%. Carroll concludes that while ‘‘… only one-half the endogenomorphic patients can be identified by this laboratory procedure and a normal suppression response to dexamethasone will not exclude a diagnosis of endogenomorphic depression which would otherwise have been made on clinical grounds’’ (3). The severity of the melancholic episode increases the likelihood of high serum cortisol levels and DST non-suppression. This is best seen in the high specificity of the abnormal DST in patients with delusional (psychotic) depression where detailed studies find high plasma and 24-h UFC levels both before and after dexamethasone in psychotic depressed patients (11-17). Psychotic schizophrenic patients typically do not show abnormal cortisol levels, arguing that the high levels are not a characteristic of the psychosis, but of mood disorder (18). A meta-analysis of 14 studies comparing DST results in psychotic and non-psychotic depressed patients found the non-suppression rate to be substantially higher in psychotic depressed patients (19). Depressed patients with abnormal DST have a higher risk for suicide (20). In one 15 year follow-up study the suicide risk was 27% compared with 3% among patients with a normal DST (21). A review of 101 patients re-examined over 2 years confirmed the higher risk for suicide and higher risk for hospitalization for suicidality in those with abnormal DST (22). Patients with abnormal cortisol metabolism are more likely to make suicide attempts (23-28). To be sure, other studies do not find the association (29-35). The DST was recognized as a state marker for depressive mood disorder. It was quickly applied as a diagnostic screening test. In patients with diagnoses other than depressive mood disorder, the incidence of abnormal DST was considerably lower. In patients with bipolar disorder, the assessments were complicated by the high incidence of depressive moods and symptoms. The frequency of non-suppression in manic patients varied from 0% to 70% with the highest frequency in mixed bipolar disorder (36-41). The development of DSM-III in 1980 offered new diagnostic criteria, ostensibly based on objective criteria. Mood disorders were divided into major depression, bipolar disorder, and dysthymia with the separations poorly defined. Historical features, not the immediate pathophysiology, determined the diagnosis. When the DST was applied across these new diagnostic classes, the correlations between test and diagnosis fell. By 1985, an extensive review concluded that: ‘‘The sensitivity of the DST in major depression is limited (about 44% in over 5000 cases) but is higher in psychotic affective disorders and mixed manic-depressive states (67–78%). The high specificity of the DST vs. control subjects (over 90%) is not maintained vs. other psychiatric disorders (77% specificity overall), and acute ‘‘distress’’ may contribute to non-suppression of cortisol’ (42). Yet, these authors conclude: ‘The test may have power in differentiating severe melancholic depression, mania, or acute psychosis from chronic psychosis (87% specificity) or dysthymia (77% specificity)’’. It was at this juncture that a committee of the American Psychiatric Association examined the DST as a laboratory test (43). Reviewing the available reports to the end of 1986, the authors concluded: ‘‘The sensitivity of the DST (rate of positive outcome, or non-suppression of cortisol) in major depression is modest (about 40–50%) but is higher in very severe, especially psychotic, affective disorders, including major depression with psychotic as well as melancholic features, mania and schizoaffective disorder’’. As to the findings in normal subjects: ‘‘The specificity (true negative outcome) of the DST in normal control subjects is above 90%, but it varies from 90% in psychiatric conditions that often need to be separated from major affective disorders’’. With regard to prediction of treatment response the committee wrote: ‘‘Positive initial DST status in major depression does not add significantly to the likelihood of antidepressant response, and a negative test is not an indication for withholding antidepressant treatment’’. Further, they noted: ‘‘Some recent data suggest that DST-positive depressions (cortisol non-suppression) are less likely than DST-negative cases (cortisol suppression) to respond to placebo’’. Finally, in considering prognosis, the committee concluded: ‘‘Failure to convert to normal suppression of cortisol with apparent recovery from depression suggests an increased risk for relapse into depression or suicidal behavior’’. They ended: ‘‘Although the clinical utility of the DST as currently understood is limited, in certain specific situations its thoughtful use may aid clinical decision making’’. In defense of the DST, Carroll responded: ‘‘Modest and reasonable proposals for the DST and sleep electroencephalogram (EEG) became exaggerated; methodological, conceptual, and interpretive caveats were ignored; and some clinicians adopted these tests uncritically with the enthusiasm of magical, absolutist thinking. … An unfortunate consequence of such positive and negative absolutist thinking has been the tendency to hold laboratory measures to a higher standard than other external validators of diagnosis’’ (44). Carroll later recalled that the greatest challenge to the test was ‘‘the arbitrary, data-free change in the clinical diagnostic criteria for depression and melancholia introduced by the American Psychiatric Association in 1980 with the DSM. This change had the effect of creating a non-validated ‘‘gold standard’’ against which the DST inevitably was compared’’ (45). The DST was developed in Australia with the ICD-9 clinical diagnostic system. In this review of his experience, Carroll also notes: ‘‘The subgroup of depressed patients with abnormal DST results resembles the classic melancholic clinical profile, has a high rate of recurrence and a strong family history, has a poor prognosis if the test does not normalize during treatment, and has the highest rate of response to antidepressant drugs. Abnormal DST results are strongly associated with suicide or violent suicide attempts. … this group of patients will fail to respond to psychosocial treatment of their depressions …’’. At the time, the standard antidepressant treatments were TCAs and ECT. After this cascade of variable assessments, and with the finality of a Supreme Court pronouncement, the DST was rejected as a clinical measure of interest. It has not been applied in the assessments of new antidepressant treatments so that these evaluations have been necessarily carried out in very heterogeneous populations of ‘depressed’ patients (46, 47). Nor is the DST applied to assess improvement or relapse. Laboratory tests were rejected in DSM (and ICD) psychiatric classification systems. This rejection is a disservice to our science. A comparison of this history with that of electroencephalography is illuminating. The scalp-recorded EEG was developed by Hans Berger in 1929 as a physiologic measure of interest in psychiatric patients. The EEG varied with age, sleep, and vigilance; sensitively changed with psychoactive drugs; reflected acute deliria and dementia; and offered unique measures of brain pathology. The EEG identified the location of mass brain lesions, especially irritative lesions. This application was has been supplanted by the CAT, magnetic resonance imaging, and other brain imaging methods. A science of sleep EEG emerged to become the present guide to the diagnosis and treatment of sleep disorders. The EEG's unique characteristic patterns in patients with seizure disorders led to a useful test in the diagnosis and management of epilepsy. The test was not considered absolute. In patients with well defined epilepsy, single EEG recordings are abnormal in 56%. Repeated second and third samples increase the incidence of abnormal recordings by 26%, for an overall success rate of 82%. On the other hand, in normal subjects, after controlling for a history of head injury, medication use, illicit substance abuse, and central nervous system infections, the incidence of dysrhythmic EEG recordings are obtained in <2% of normal subjects (48-50). In the ensuing decades, the variations because of age and medication, the merits of different activation procedures (hyperventilation, photic stimulation, sleep), the utility of 24-h video-EEG recording, improvements in statistical processing of the data, and the testing of different classifications improved the reliability of EEG diagnosis of seizure disorders to a science (51-53). Yuuki et al. (4) show us that the DST, even with its limitations, is a useful index of the presence and severity of mood disorder and the change with treatment. The DST has the potential to be as reliable and as useful in assessing mood disorders as the EEG is in seizure disorders. Its rejection for a quarter of a century and the profession's failure to devise any more reliable measure has left psychiatric diagnosis of mood disorders in shambles. The rejection of laboratory tests by DSM classification commissions was in my opinion an error. Neuroendocrine tests define a characteristic population of depressed patients best labeled ‘melancholic’. They offer a way to insure greater reliability in diagnosis and greater homogeneity in populations assessing psychiatric treatments. It is time that the profession re-examines the DST and the DEX/DST in clinical psychiatric practice.