- Email: [email protected]
Plasma dexamethasone concentrations and the dexamethasone suppression test
BIOL PSYCHIATRY 1990;27:159-173
! 59
Plasma Dexamethasone Concentrations and the Dexamethasone Suppression Test James C. Ritchie, Beth M. Belkin, K. Ranga R. Krishnan, Charles B. Nemeroff, and Bernard J. Carroll
Altered bioavailability or altered pharmacokinetics of dexamethasone (dex) may contribute to a positive Dexamethasone Suppression Test (DST) in psychiatric patients. We measured plasma dex and plasma cortisol concentrations in 32 patients with primary major depressive disorder (MDD), 14 patients with other psychiatric disorders, and 16 normal controls. Cortisol was measured by the competitive protein binding (CPB) assay and dex by RIA (lgG Corp.). AdditionaUy, cortisol was measured by a fluorescent polarization immunoassay (FPIA) available on the Abbou TDx analyzer in an attempt to validate this method for use in the DST. The agreement between FPIA and CPB cortisol results was excellent. Depressed nonsuppressors, by definition, had significantly higher mean plasma cortisol concentrations than depressed suppressors, psychiatric controls~ and normal volunteers at 8:00 Ata, 3:00 eta, and 10:00 eta postdex. When DST nonsuppressors and suppressors were compared regardless of diagnostic group, plasma dex concentrations were significantly lower (p < 0.01) in the DST nonsuppressors. There was a significant negative correlation ~etween plasma cortisol levels and plasma dex levels across all subjects at 8:00 ~¢ (r = -0.365, n = 44, p < 0.05). When the subjects were sorted by diagnostic category, there was a strong, but not statistically significant, trend toward lower plasma dex concentrations in the melancholic nonsuppressors versus the melancholic suppressors and between the psychiatric control non-supppressors and the corresponding suppressor group. These relationships disappeared when we restricted our analyses to an empirically derived middle range of plasma dex concentrations within which the DST results were considered to be valid. We conclude that bioavailability or pharmacokinetics of dex may significantly contribute to DST results. Further investigation is needed to determine whether or not the quantification of dex and its metabolites and their determination at which specific timepoints during the DST will enhance the predictive or interpretive value of the DST in psychiatric patients.
Introduction Patients with endogenous depression (ED) frequently have abnormalities of the hypothalamic-pituitary-adrenal (HPA) axis that resu!t in disinhibited cortisol secretion. Findings
From the Departments of Psychiatry (J.C.R., B.M.B., K.R.R.K., C.B.N., B.J.C.) and Pharmacology (C.B.N.), Duke University, Durham, NC. Supported in part by NIMH Grants MH 39593, MH 40159, and MH 42510. Address reprint requests to Dr. J. C. Ritchie, Department of Psychiatry, Duke University, Durham, NC 27710. Received February 16, 1988; revised January 23, 1989. © 1990 Society of Biological Psychiatry
0006-3223/90/$03.50
160
BIOL PSYCHIATRY 1990;27:159-173
J.C. Ritchie et al.
in such patients include hypersecretion of cortisol, inappropriate nocturnal secretion of cortisol, and escape from suppression of cortisol during the Dexamethasone Suppression Test (DST) (Carroll et al. 1976, 1980a, 1981). Among properly screened psychiatric patients, this last finding has been suggested to be specific for the melancholia (Carroll et al. 1981). It is believed to be due to a functional disturbance of the limbic system, which results in a limbic-hypothalamic overdrive, leading ultimately to hypersecretion of corticotropin-releasing factor (CRF), corticotropia (ACTH), and overproduction of cortisol. Elevated CRF concentrations have, in fact, been demonstrated in the cerebrospinal fluid (CSF) of depressed patients (Nemeroff et al. 1984; Banki et al. 1987), though it is not certain that the CSF concentration of CRF reflects the activity of the HPA axis. This hypothesis implies that there is a central nervous system (CNS) dysfunction in depression and rests on the assumptions that (1) cortisol secretion in both DST suppressors and nonsuppressors parallels that of ACTH, and (2) peripheral abnormalities, such as altered absorption or metabolism of dexamethasone, do not account for the abnormal DST results. Studies of plasma dexamethasone (dex) concentrations in DST suppressors versus DST nonsuppressors have yielded inconsistent findings. There have been several reports of significant negative correlations between postdex plasma cortisol concentrations and plasma dex concentrations (Arana et al. 1984; Holsboer et al. 1983, 1984, 1986; Morris et al. 1986; Johnson et al. 1987). In contrast, a number of investigators have been unable to find an association between plasma dexamethasone concentrations and DST response. For example, Carroll et al. (1980) found mean 4:00 PM dexamethasone concentrations of 0.75 ng/ml in 19 abnormal tests and 0.72 ng/ml in 26 normal tests (using a plasma cortisol criterion of 6 Ixg/dl). Low dexamethasone concentrations (<1.0 ng/ml) were found in 14 of the 19 abnormal tests (74%) and in 20 of the 26 normal test (77%). Rubin et al. (1980) similarly found no significant difference in a smaller early study. More recently, Poland et al. (1987) reported no significant differences in plasma dexamethasone cencentration between depressed patients with and without abnormal DST results. On the other hand, control subjects with abnormal DST results did have significantly lower plasma ,texamethasone concentrations than those with normal suppression. Discrepancies in the literature may be due to methodological factors, such as subject selection, sampling times, differences in assay sensitivity and specificity, and possible artifacts associated with uncharacterized dex metabolites and the use of unextracted plasma in the radioimmunoassays (RIAs) for plasma dexamethasone. The BST for use in psychiatry was standardized using a modified version of .k.,~.,~ competitive protein binding method for plasma cortisol (Murphy 1967). This method, developed in the late 1960s, measures total glucocorticoids and is particularly suited to accurately measure the relatively low levels encountered in the DST. The cortisol method used in most clinical laboratories, however is RIA. The disparities in these methods and the resultant use of different cortisol concentrations as the positive/negative DST cutoff point have contributed greatly to the contr,,, :rsy surrounding the DST. Previously (Ritehie eta!. 1985), we evaluated several RIA methods of plasma cortisol determination for use in the DST. Recently, a nonisotopic immunoassay for cortisol has been introduced. This method, available on the Abbott TDx, utilizes fluorescent polarization immunoassay (FPIA) technology. It is highly specific and sensitive for cortisol. Additionally, the method requires no pretreatment of the sample, is automated, and does not use radioactive isotopes (eliminating health hazards and waste disposal concerns). The present study thus was undertaken to examine further the relationship of plasma dex (utilizing the only commercial assay method available) to plasma cortisol (utilizing
Plasma Dexamethasone and DST
BIOL PSYCHIATRY 1990;27:159-173
161
the assay methodology originally employed to standardize the DST) in a rigidly defined group of depressed inpatients, nondepressed psychia~c inpatients, and normals. Aaoitionally, we sought to validate a new fluorescent polarization immunoassay of plasma cortisol for use in the DST. Methods
Subjects We studied 32 patients with major depressive disorder (MDD), primary endogenous subtype, who were hospitalized on the Affective Disorders Unit, Depart,merit of Psychiatry, Duke University Medical Center (DUMC). The diagnosis of MDD was established on the basis of a structured interview, the Schedule for Affective Disorders and Schizophrenia (SADS) (Spitzer and Endicott 1977), by a member of the research team; unstructured clinical interviews by a senior psychiatrist; review of past psychiatric history; and a comprehensive medical and clinical chemistry assessment to screen for medical disorders. Following this evaluation, patients were assigned a consensus R~search Diagnostic Criteria (RDC) (Spitzer et al. 1978) diagnosis and a (clinical) DSM-III d:..gnosis. Patients were included only when there was agreement among the staff about the diagnosis. We also studied 14 patients with other p~ychiat~ic disorders. This group included schizophrenic patients at John Umstead Hospital, Burner NC (n = 11) and patients being evaluated for chronic pain on the Clinical Specialities Unit (CSU) at DUMC (n = 3). To avoid any potentially confounding effects of the acute stress of hospitalization, the DST was never performed prior to the fourth day after admission. Patients were excluded if they met the medical, drug, or technical exclusion criteria described by Carroll et al. (1981). Finally, 16 normal volunteers, screened for medical disorders and psychiatric illness and without a family history of psychiatric illness, were studied. All subjects received an oral dose of 1.0 mg dexamethasone at 11:00 PM, and blood samples were obtained from inpatients on the following day at 8:00 AM, 3:00 PM, and 10:00 PM. Only the 8:00 AM and 3:00 PM blood samples were obtained from the normal controls. Nonsuppressor (NS) status was defined by any postdex plasma cortisol concentration > 5 ~g/dl (Carroll et al. 1981).
Assays Blood was collected in hepannized tubes for measurement of cortisol and dex. The plasma was separated from whole blood within 15 min, aliquoted in 0.5-ml portions, and frozen at -80°C until the analyses were performed. Cortisol was measured by a modified competitive protein binding (CPB) assay (Murphy 1967). It is well established that dex does not bind to plasma corticosteroid binding globulin (CBG: transcortin), and therefore, dex does not interfere with the CPB assay for cortisol. This procedure has been used in our laboratory for over 10 years, and coefficients of variation (CV) have been 12.6% (interassay) and 6.4% (intraassay) for a cortisol-poor sample (2.6 gg/dl) and 8.9% (interassay) and 5.6% (intraassay) for a cortisol-rich sample (10.15 ~g/dl) over the last 108 assays. We evaluated the Abbott TDx cortisol assay using the same paradigm we used previously to compare RIA methods for plasma cortisgl (Ritchie et al. 1985). We used the same five postdexamethasone pools as in our previous study. The pools were analyzed
162
BIOL PSYCHIATRY 1990;27:159-173
J.C. Ritchie et al.
in replicates of 10 on the TDx utilizing exactly the procedure supplied by the manufacturer. The pools were also analyzed in 19 replicate competitive protein binding assays. The results from the two procedures were then compared and a tentative cutoff value for the DST, utilizing the TDx methodology, was derived. To further validate the Abbott FPIA method and our tentative DST cutoff, we determined the plasma cortisol concentrations on the samples obtained from the patients and normal volunteers described above. Using these results, we then compared the cutoff points for a positive DST in the two procedures. The antiserum (IgG-dexamethasone-1) and protocol for the dex assay were supplied by lgG Corporation (Nashville, TN). The procedure was used with one modification. In an effort to improve the sensitivity of the assay, we used 1,2,4,6,7-3H-dexamethasone (New England Nuclear, Boston, MA) as the tracer, rather than the two prime dexamethasone trace specified by IgG Corp. The assay buffer was 0.063 M sodium dibasic p~,osphate, 0.013 M EDTA, 0.02% sodium azide. This was adjusted to pH 7.4 with HC1, and then 1 mg/ml of bovine ~amma globulin and 8-anilino-l-napthalenesulfonic acid (magnesium salt) were added to decrease nonspecific binding. The primary antiserum for the assay was prepared in rabbits against a 3-oxime conjugate of dexamethasone. For use in the assay, it was diluted 2500-fold in assay buffer plus 1% (v/v) normal rabbit serum. Standards were prepared fresh for each run in assay buffer. The standard curve was run from 0.025 ng/ml to 10 ng/ml, in triplicate. The assay is run on an unextracted plasma. For assay, 0.1 ml of standard or plasma is incubated with 0.1 ml of diluted primary antiserum and 0.1 ml of 3H-dex tracer (approximately 5500 dpm) for 20 hr at 4°C. Twenty-five microliters of a second antibody, purified goat antirabbit gamma globulin (Arnel Products, NY) is then added, mixed, and incubated for 2 hr at 4°C. Separation buffer (0.063 M sodium phosphate, 0.013 M EDTA, 0.02% sodium azide, 2.5% bovine serum albumin, pH 7.4), is added to each tube (1.6 ml). The tubes are centrifuged at 6000 x g and 4°C for 30 min. The supernatants are decanted and allowed to drain for 3 min. Then, 0.4 ml of NCS tissue solubilizer (Amersham Corp., Arlington Heights, IL) and 50 I~1 of 25% acetic acid are then added to each tube. Scintillation cocktail (2 ml) is added to each tube. The tubes are shaken and then counted on a model 460 Packard Instr. scintillation counter (Downer's Grove, IL). The scintillation counter automatically corrects for quench and luminescence. The resulting data are then reduced using standard RIA log/logit functions. The minimal detectable quantity of dex in the assay is 0.02 ng/ml. Over the last 30 assays, the 80%, 50%, and 20% binding points have averaged 0.07, 0.33, and 1.46 ng/ml, respectively. The interassay CV is 8% at a level of 0.52 ng/ml, and the intraassay CV is 4% at the same level. Quality control pools, prepared from dex and prednisolone-suppressed normals, are run in triplicate in every assay. Cross-reactivities with other major steroids, as stated by the manufacturer and verified in our laboratory, are all less than 1%. The cross-reactivities of dex metabolites are unknown.
Statistics Both the postdex plasma cortisol concentrations (determined by the CPB methodology) and the plasma dex concentrations were log-transformed because of inhomogeneity of variance. Mean plasma cortisol and dex concentrations of MDD suppressors and MDD nonsuppressors were compared utilizing the two-tailed Student's t-test for unmatched samples. Means among the four groups (MDD suppressors, MDD nonsuppressors, psychiatrically ill others, and normal controls) were compared utilizing one-way Analysis
Plasma De~methasone and DST
BIOL PSYCHIATRY 1990;27:159- !73
163
Table 1. Plasma Cortisol and Dexamethasone Concentrations during the DST Irrespective of Subject Groupsa Suppressors
Nonsuppressors
Time
Mean cortisol (ttg/dl)
Cortisol SD range
Mean dex (ttg/ml)
Dex SD range
Mean cortisol (~g/dl)
Cortisol SD range
Mean dex (ttg/ml)
Dex SD range
8:00 AM 3"00 PM 10:00 PM
1.06 1.14 1.17
0.5--2.1 0.6--2.2 0.5--2.5
2.25 0.91 0.40
1.19-4.24 0.4-1.8 0.2--1.04
6.78 6.22 6.97
2.25-18.5 2.0-19.0 2.9-16.6
1.24 0.54 0.19
0.8-2.0 0.2-1.2 0.06-0.6
aAll data are log-transformed. The p-vah~es for the cortisol comparisons were all <0.001. For dexamethasone, the p-values were <0.005~ <0.005, and <0.01 for the respective timepoints.
of Variance (ANOVA), followed by a multiple comparisons test, either the Newman Keuls pairwise procedure or the Scheff6 procedure. Relationships among the variables were evaluated utilizing Pearson correlations. The terminal (beta-phase) half-life of dex for each group was calculated using standard first-order kinetics (Kalent et al. 1985). The power of the analysis was determined by the method of Dixon and Massey (1969) for two-sided Analysis of Variance, utilizing the power curves in the same work (p 524). The two cortisol assays were evaluated utilizing least-squares linear regression and Pearson correlations. Goodness of fit was assessed by analysis of the residuals of the regression and t-test. Results The average age of the subjects was 45.4 _+ 16.7 years. There were no significant differences in age among the groups. The depressed group had a mean age of 52.9 _+ 15.5 years, the psychiatric controls had a mean age of 36.5 _+ 14.1 years, and the normal controls had a mean age of 34.3 _+ 10.98 years. There were no significant correlations between age and plasma alex or postdex cortisol concentrations. There were 36 men and 26 women overall in our population. In the depressed group, there were 19 men (9 NS) and 13 women (9 NS). The psychiatric controls were composed of 10 men (5 NS) and 4 women (0 NS). The normal control group consisted of 7 men (0 NS) and 9 (0 NS). There were no significant correlations between sex and plasma dex or postdex plasma cortisol concentrations. Within the total group of patients tested, 23 were nonsuppressors and 21 were suppressors. Of the 32 patients with MDD, 18 (56%) failed to suppress, whereas 5 (36%) of the 14 patients with other psychiatric disorders were nonsuppressors. None of the normal volunteers escaped dex suppression. The means of the log-transformed plasma cortisol and dex concentrations at each timepoint of the DST for the nonsuppressors and suppressors irrespective of diagnosis are presented in Table I. As expected, the mean plasma cortisol levels for the nonsup. pressors were significantly higher at each time of sampling than those of the suppressor~,i. In contrast, the mean plasma dex levels of the nonsuppressors were significantly lower at each time than those of the suppressors. When subjects were examined with respect to diagnostic status, depressed non~uppressors and nonsuppressing psychiatric controls again had significantly higher mean (logtransformed) plasma cortisol concentrations (p < O.Ol) than depressed suppressors, psy-
164
J.C. Ritchie et al.
BIOL PSYCHIATRY 1990;27:159-173
Table 2. Mean Plasma Cortisol and Dexamethasone Concentrations during the DST by Subject Groups° Nonsuppressors
MDD (n = 18)
Psychiatric controls (n = 5)
8:00 AM 3:00 PM 10100 PM
3.34 9.79 10.17
4.43 6.87 9.19
8:00 AM 3:00 PM 10:00 PM
1.40 0.77 0.31
1.34 0.56 0.24
Suppressors
Normals (n = 0)
MDD (n = 14)
Psychiatric controls (n = 9)
Normals (n = 16)
1.35 1.01 1.20
1.24 1.41 1.28
2.05 0.81 0.48
3.20 I. 14 0.39
Plasma cortisol (p.g/ml) -~ --
1.32 1.64 1.88
Plasma dexamethasone (izglml) ----
2.69 1.50 0.69
aAll data are log-transformed.
chiatric control suppressors, and normal volunteers at all three timepoints. There was a strong, but not statistically significant trend, toward lower mean plasma dex concentrations in MDD-NS versus MDD-S and between psychiatric control-NS versus psychiatric control-S subjects at all sampling times. These data are shown in Table 2. Figures 1, 2, and 3 are scattergrams of the log-transformed cortisol values plotted against the log-transformed dex values for each timepoint of the DST. The individual data points are coded by subject category. Suppression status is in regard to the overall DST and not necessarily the individual timepoint displayed in each figure. There was a significant inverse relationship between the plasma cortisol levels and plasma dex levels across all subjects irrespective of diagnosis at 8:00 AM (r = - - 0 . 3 6 0 ,
Figure 1. Log cortisol vs log dexamethasone concentrations at 8:00 AMtimepoint of DST.
Pldsma Dexamethasone and DST
BIOL PSYCHIATRY 1990;27:159-173
165
Figure 2. Log cortisol vs. log dexamethasone concentrations at the 3:00 PM timepoint of DST. n - 45, p < 0.025), 3:00 PM (r -- --0.380, n = 60, p < 0.005), and 11:00 PM (r = --0.400, n = 55, p < 0.005). However, tiffs relationship was not maintained when the groups were sorted by diagnoses, with the exception of the psychiatric controls. In this group, a significant negative correlation between plasma control cortisol and dex values was observed, but only at the 8:00 AM timepoint (r - - 0 . 6 3 3 , n - 11, p < 0.05).
Figure 3. Log cortisol vs. log dexamethasone concentrations at the 10:00 PM timepoint of DST.
166
J.C. Ritchie et al.
BIOL PSYCHIATRY 1990;27:i59-i73
Table 3. Plasma Dex Concentrations and the Dex Validity Windows Nonsuppressors
Suppressors
Time
Dex window (l~g/ml)
n
Mean
n
Mean
8"00 AM 3:00 PM 10:00 PM
0.96-2.31 0.34-1.67 0.10-0.91
8 12 15
1.60 0.93 0.33
15 31 27
1.47 0.93 0.38
If the analysis of these data are restricted to only those samples with midrange dex concentrations (established empirically by inspection of the distributions) at each time point (8:00 AM, 0.96-2.31 ng/ml; 3:00 PM, 0.34-1.67 ng/ml; 10:00 PM, 0.1--0.91 ng/ml), the significant difference in plasma dex concentrations between suppressors and nonsuppressors, irrespectiveof diagnosis, is lost. These data are shown in Table 3. Additionally, the observed inverse association between plasma cortisol concentrations and plasma dex concentrations at each timepoint across all subjects, irrespective of diagnosis, disappears (p > 0.1 for all the correlations). We used the plasma dex concentrations obtained in this study to calculate the elimination halfAife (t~]2) for dex in each group. These calculations assume first-order kinetics and appear to be justified based on the position of the three DST samples late in the terminal elimination phase of dex. The mean t 1/2 of the normal controls was 4.61 hr. Overall, in the patient groups, both nonsuppressors (t ~/2 5.62 hr) and suppressors (t !12 6.11 hr) had half-lives greater than the normals. When broken down by patient category, this increase in t ~/z persisted in all groups. In the depressed patients, nonsuppressors had a t ~/2of 6.43 hr, whereas the suppressors had a t ~]2of 7.13 hr. In the psychiatric controls, the nonsuppressors had a t ~2 of 5.64 hr and suppressors 6.68 hr. None of these comparisons reached significanc¢~, due to the large variability exhibited by the subjects. However, we aid note a trentl (p = 0.06) for larger t~/2s in the patient suppressor group when compared to the normal controls (6.90 hr versus 4.61 hr). The results for the comparison of the two eortisol assay methods utilizing the postdexamethasone plasma pools were excellent (r = 0.9985, slope = 0.88, y-intercept = 0.90). The results for the fifth pool were not used, as this pool was beyond the range of linearity for the CPB procedure, and thus, direct comparisons could not be made. The 95% confidence interval for a TDx value, which corresponded to a CPB value of 5.0, was 4,88-5.72 p,g/dl. Based on these results, we tentatively set the cutoff value for the DST utilizing TDx cortisols at 5.0 I~g/dl. We then determined the plasma cortisol concentrations on the 169 postdexamethasone samples generated in this study, in duplicate, on the TDx and in the CPB assaY. These results are shown in Figure 4. The procedures were highly correlated, and when mean TDx values were compared to mean CPB values, they yielded a correlation coefficient of 0.9942 (slope = 0.89, y-intercept = -0.08). These results show a significant linear relationship between the two procedures, with a p-value of < 0.01. As more than half of the values in the above comparisons were <5 I~g/dl, we also plotted the residuals from the regression analysis. Though not shown, this plot showed no consistent error trend along the regression line. The determination of accurate levels of cortisol between 3 and 8 I~g/dl is critical for the DST. Figure 5 presents such a subset of the data from Figure 4. In Figure 5, we replotted those samples that had CPB cortisol values >3 and <8 ~g/dl. As can be seen, the two procedures are still highly correlated, with the exception of one obvious outlier that had a CPB value of 8 and a TDx value of 2 ~g/dl.
Plasma Dexamethasone and DST
BIOL PSYCHIATRY 1990;27:159-173
!~_.POSI-II~ R~tA Stm
,° I 35
r -- 0 . 9 9 4 2
-
SLOPE = 0 . 8 9 Ycept = - 0 . 0 8
30
n = 169 25
a
..I
0 ¢n 2 0 I-., I]=: 0 15 ¢J, o I-
q a
o
o
•
Q
10
aa
5
oJ
P"
13
uO/dL
I
I
O
5
.
I
.
I
10
,
15
I
.
I
20
,
I
25
.
30
I
,
35
I._
40
C.P.B.L. CORTISOL CPB Figure 4. TDx cortisol vs. CPB cortisol assay; 169 postdex plasma samples.
PLASMA SAXPLES > 3 a n d
<8
ug/dL
10 r =
0.7376
SLOPE = Ycept =
0.73 -0.14
n=31 -J 0 fJ) I0=: 0 ¢J ,ell
0
°a ~
a
N
a aaaaa
I-
I
0
I
I
2
,
_
I _.
4
,
I
.
6
I
.I
ug/dL
8
G.P,B.L. CORTISOL CPB Figure 5. TDx cortisol vs. CPB cortisol assay; plasma samples >3 and <8 I~g/dl.
I..
10
167
168
J.C. Ritchie et al.
BIOL PSYCHIATRY 1990;27:159-173
Table 4. Comparison of CPB and TDx Cortisol Values in the DST: Effect of Varying TDx Cutoff on Test Outcome TDx cutoff point 3.0 3.5 4.0 4.5 5.0 5.5 6.0
Izg/dl Izg/dl Ixg/dl p.g/dl p,g/dl p,g/dl ~g/dl
CPB + TDx +
CPB TDx +
CPB + TDx -
43 43 41 40 36 34 29
8 4 2 1 1 1 0
1 1 3 4 8 10 15
CPB TDx 117 121 123 124 124 124 125
Percent overall agreement° 94.67 97.04 97.04 97.04 94.67 93.49 91.12
°Using the CPB assay and a DST cutoff of 5.0 Izg/dl, there were 44 positive and 125 negative results.
Discussion The purpose of the comparison of the two cortisol assay methodologies in this study was to determine a DST cutoff, using the TDx assay, that best compared to the standardized DST cutoff, using the CPB assay. We made an initial evaluation of the two methods utilizing the postdex cortisol pools that we have used previously (Ritchie et al. 1985) i~,r this purpose. This evaluation showed the methods to be comparable, and we determined that a TDx DST cutoff of 5 ~g/dl was ~ost appropriate. We next ran parallel studies on the 169 postdex plasma samples generated in this study. This was done to further optimize the DST cutoff in the TDx assay by using real patient samples. These results are summarized in Table 4. As can be seen from the table, all of the possible TDx cu'~erf points, from 3 to 6 ~g/dl, gave an overall agreement of >90% with the CPB cutoff of 5 ~g/dl. We therefore recommend that a value of 5 Izg/dl be used as the cutoff criteria for the DST when using either the TDx or CPB method of cortisol analysis. Also, as per our previous recommendation, we continue to believe that all postdexamethasone cortisol concentrations >3 and <7 ~g/dl should be confirmed by reanalysis of the samples. This is due to the importance of a positive test result to the psychiatrist and does not reflect any inaccuracy in either measurement technique. In this study, we have attempted to define the relationship of postdex plasma cortisol concentrations to circulating dexamethasone concentrations during the l-mg DST. We used the cortisol assay methodology originally used to standardize the DST and the only commercially available plasma dex RIA in an attempt to establish this relationship (plasma dex to plasma cortisol) in the most reliable and easily available methodologies. Our results demonstrate considerable individual variability in plasma dex concentrations following a 1-mg oral dose in depressed patients, psychiatrically ill others, and normal controls. These findings are consistent with those of previous investigations. In fact, a 100-fold variation in 4:00 PM plasma dexamethasone concentrations (range 0.1-9.1 ng/ml) can be seen in the combined results of patients and control subjects reported to date (Arana et al. 1984; Carroll et al. 1980; Holsboer et al. 1984, Johnson et al. 1984; Morris et al. 1986; Poland et al. 1987). Most l:revious studies have attributed this broad range to significant individual variations in dex metabolism. An alternative hypothesis, however, is that this broad range may be due to assay interference by unknown dex metabolites. Further investigation will be needed to clarify this point. When we examined the entire range of plasma dex concentrations in all subjects irrespective of diagnosis, we found a significant inverse relationship between dex levels
Plasma Dexamethasone and DST
BIOL PSYCHIATRY 1990;27:159-173
169
and plasma cortisol concentrations. However, we found no significant difference in plasma dex concentrations between DST suppressors and nonsuppressors when only patients with diagnosis of MDD were considered, though a strong trend in this direction was evident (Table 2). Post hoc power analysis (Dixon andMassey 1969) yielded power factors of >0.7, >0.6, and >0.6 at the three respective timepoints of the DST for the comparison of plasma dex concentrations between MDD-suppressors and MDD-nonsuppressors (based on an oLof 0.05 and 14 individuals per group). The analysis also revealed that given the large variance displayed by each group, we would have to increase the size of each group by a factor of 5 to reach a power factor of 0.8. These results are consistent with those of several studies that found a negative correlation between plasma cortisol and plasma dex levels in psychiatric patients and normal controls. Morris et al. (1986) also observed wide individual variation in serum rex concentrations following a !-mg dose in 51 patients representing a cross-section of all psychiatric disorders. They, too, found a significant inverse relationship between serum dex levels and serum cortisol levels. Johnson et al. (1984) observed considerable individual variation in plasma dex levels across patients and normal subjects and noted an inverse relationship between serum dex and cortisol levels. Studies of MDD suppressors versus MDD nonsuppressors are conflicting as to differences in plasma dex levels. In the first study of plasma dex concentrations during the DST in psychiatric patients, Carroll et al. (1980b) measured the plasma dex levels in 4 endogenously depressed patients before treatment (when abnormal DST results were observed) and again after recovery (when the DST results had normalized). No differences in plasma rex concentrations or t~/2s were observed at any timepoint. They also compared plasma dex concentrations in 10 endogenous depressed, age- and gender-matched subjects with 10 nonendogenous depressed outpatients. All of the former had abnormal DST results, and all of the latter suppressed normally. No differences in plasma dexamethasone concentrations were found. Also, in 1980, Rubin et al. (1980) studied 15 patients with MDD and 8 normal controls matched to the patients on age, gender, race, and (for women) endocrine status. They found no difference in 7:00 AM postdex serum dex ievels or in average dex half-lives among suppressors, nonsuppressors, and normal controls. In a recent study, Poland et al. (1987) found no significant dexamethasone concentration differences in MDD patients (15 DST nonsuppressors and 25 suppressors). Although they observed a trend toward reduced serum dex concentrations and an inverse, correlation between postdex serum cortisol and dex levels in the nonsuppressing patients, nei,'her of these effects reached statistical significance. The first finding of plasma dexamethasone concentration difference between MDD suppressors and MDD nonsuppressors was reported by Holsboer (1983). Another group reported similar results in later studies (Johnson et al, 1984, 1987). In another recent report, Maguire et al. (1987) described a bimodal distribution of dexamethasone concentrations among MDD nonsuppressors: one subgroup had low levels similar to those of volunteer subjects with abnormal DST results, whereas the second subgroup of MDD nonsuppressors had dexamethasone levels equivalent to those of normally suppressing volunteer subjects. The original finding of a significant difference in dex concentrations within MDD patients (i.e., MDD suppressors versus MDD nonsuppressors) was demonstrated again by Holsboer et al. (1984). In this study, these investigators found significant concentration differences in 16 MDD nonsuppressors when compared with 6 MDD suppressors and 6 normal controls at one timepoint during the DST. In a more recent study (Holsboer et
170
BIOL PSYCHIATRY 1990:27:159-173
J.C. Ritchie et al.
al. 1986), this group used the test-retest paradigm to examine the plasma dex concentrations of 31 MDD patients whose DST results converted with treatment. They found significantly lower mean plasma dex concentrations during illness than after recovery. Additionally, using frequent catheter sampling, they observed a decreased t ~/2 (160 min) in 5 depressed patients with abnormal DSTs, compared with 6 other depressed patients whose DSTs were normal (422 min). In 14 other patients with MDD who exhibited adequate DST suppression during the depressive state and after recovery, the dex concentrations were unchanged. Our results indicated a trend toward a significant plasma dexamethasone difference between MDD suppressors and MDD nonsuppressors. That this trend was not statistically significant may be due to the small sample size. It may also be explained by differences in the dex assay methodologies used. For example, in 4 separate reports, the 4:00 PM mean plasma dexamethasone concentration associated with abnormal DST results ranged from 0.50 to 2.18 ng/ml (Arana et al. 1984; Holsboer et al. 1984; Johnson et al. 1985, this study, Table 2). This variation of 2-4-fold across studies strongly suggests that the dexamethasone assay methods reported from different laboratories are not equivalent. It could be that dexamethasone metabolites (so far uncharacterized) may cross-react in some assays, falsely elevating the levels in some patients. This would also explain our inability to show a difference in the half-lives between suppressors and nonsuppressors, as the German group did. Interestingly, when compared to normals (4.61 hr), all of our ho~ pitalized subjects (depressed 6.71 and other psychiatric patients 6.35) exhibited a somewhat lengthened elimination half-life. The Holsboer group did not publish any data on normals in their 1986 study (Holsboer et al. 1986). Previously reported dex elimination half-lives have varied in normals from 2.78 to 6.13 hr in a specifically designed pharmacokinetic study (Poland et al. 1987). Thus, a high level of interindividual variation seems to exist in the metabolic processing of dex, even in normal individuals. On inspection of the wide range of plasma dex concentrations in our sample, we identified a middle range of dex values below which subjects tended to escape from DST suppression and above which they never failed to suppress. We propose that those extreme values of the dex distribution represent areas of increased likelihood of false-positive and false-negative results, respectively. When we restricted our analysis to this empirically identified middle range of dex levels, we found no difference in mean plasma dex concentrations between suppressors and nonsuppressors, irrespective of diagnosis. Similarly, the previously observed inverse association between plasma cortisol and dex levels also disappeared. These findings support the hypothesis that there is a range of plasma dex concentrations within which the level of exogenous steroid attained is sufficient to suppress cortisol production in patients with normal HPA axis function, but which is not high enough to cause suppression in all patients with altered HPA axis function. We term this range the dex "window" and propose that the DST may not be valid in patients whose plasma dex values fall outside of this window. Table 3 presents our first approximation of these dex windows based on the data generated in this study. The table shows the mean plasma dex levels within these windows for suppressors and nonsuppressors, including all diagnoses. As c~n be seen, the significant difference initially observed between suppressors and nonsuppressors has disappeared. Moreover, within these windows, there was no significant correlation between plasma cortisol and dex concentrations either across all subjects or within any of tile groups at any of the three timepoints. When these validity windows are applied to our study population, 9 of 14 psychiatric controls are eliminated (4 NS and 5S), reducing the rate of nonsuppression from 36% to 20% in this group. In
Plasma'Dexamethasone and DST
BIOL PSYCHIATRY 1990;27:159-173
171
the depressed group, 11 patients are eliminated (6 NS and 5 S), and the rate of nonsuppression remains relatively unchanged at 57%. Johnson et al. (1987) were the first to propose a window concept for valid dexamethasone suppression testing in psychiatry. They studied a sample of 9 MDD, 8 psychiatric controls, and 9 normals using a 0.5-mg and 1.0-mg repeat DST paradigm and a different dex RIA than ours. In describing their results, they proposed a dex window of 1.964.32 ng/ml at 8:00 AM and 0.78-1.57 ng/ml at 4:00 PM. It is unclear whether the differences in the windows proposed by Johnson and those derived in this study (Table 3) are due to subject differences or variations in assay methodologies. Johnson et al. also noted that within the window, "significant differences in plasma dexamethasone concentrations between suppressors and nonst:ppressors are no longer evident," a finding which our data confirm. The definitive determination of window limits will require additional studies with a much larger sample size and a standardized dexamethasone assay methodology. Future studies should seek to determine the relative contribution, if any, of age, gender, weight, and hospitalization on the metabolism of dex. Additionally, studies should be pursued that seek to elucidate more fully the metabolites of dex, their cross-reactivity in the currently available assay methodologies, their bioactivity in suppressing the HPA axis, and whether or not the hypercortisolemia, which occurs in melancholia patients, induces a more rapid clearance of dex. In this regard, the recent study by Wiedemann and Holsboer (1987) is of great interest. Their data (Table 3, p 1345) strongly suggest that dexamethasone metabolites cress-react in the RIA procedure for plasma "dexamethasone," as the apparent dexamethasone concentrations were substantially higher from 2:00 AM through 11:00 PM in normally suppressing patients after oral dex administration than after intravenous dexamethasone, despite the fact that identical peak dexamethasone levels were achieved by both routes 1 hr after drug administration. In the same study, other data would be consistent with the possibility that nonsuppressing patients have accelerated dexamethasone clearance, possibly with a different metabolic profile than suppressing patients, especially after oral dexamethasone administration. Resolution of these discrepancies awaits the complete characterization of dex metabolism in normals and depressives. This in turn awaits the development of assay methodologies capable of accurately measuring parent dex and its metabolites during the standard DST. Finally, we emphasize that our plasma dex data refer only to the time period 9-23 hr after dex administration. Holsboer et al. (1986b) have shown that plasma dex concentrations during this time interval are not strongly related to the concentrations earlier in the night, 0-4 hr after dex administration, when the steroid is believed to exert its major pharmacodynamic action in suppressing the "circadian program" of HPA activity (see Carroll et al. 1976a). Additional studies, therefore, will be required to determine the optimal sampling times for relating dex pharmacokinetics to HPA suppression. Until this question is resolved, however, our preselit results, together with those of Johnson et al. (1987), suggest that the concurrent measurement of plasma dex and plasma cortisol at the standard DST sampling times (Carroll et al. 1981) can improve the clinical information provided by this test. We gratefully acknowledge the efforts of Mrs. A.C. Laws and Mrs. T. Page in the preparation of this manuscript. We also wish to thank Ms. E. Kegelmeyer for her expert technical assistance.
172
BIOLPSYCHIATRY
J.C. Ritchie et al.
1990;27:159-173
References American Psychiatric As~ ~ciation (1980): DSM-III: Diagnostic and Statistical Manual of Mental Disorders (ed 3). Wa~ingten, DC: American Psychiatric Association. Arana GW, Workman RJ, Baldessarini RJ (1984): Association between low plasma levels of dexamethasone and elevated levels of cortisol in psychiatric patients given dexamethasone. Am J Psychiatry 141:1619-1620. Banki CM, Bissette G, Arato M, O'Connor L, Nemeroff CB (1987): CSF corticotropin-releasing factor-like immunoreactivity in depression and schizophrenia. Am J Psychiatry 144:873-877. Can" B, Morris H, GiUiard L (1986): The effect of serum dexamethasone concentrations on the Dexamethasone Suppression Test. Biol Psychiatry 21:735-743. Carroll BJ (1983): The Dexamethasone Suppression Test for depression. In Habig RL (ed), The Brain, Biochemistry, and Behavior. American Association for Clinical Chemistry Publications, pp 69-84. Carroll BJ (1984): The Dexamethasone Suppression Test. In Hall RCW, Beresford TP (eds), Handbook of Psychiatric Diagnostic Procedures. New York: Spectrum Publications, pp 3-18. Carroll BJ, Curtis GC, Mendels J (1976a): Neuroendoerine regulation in depression I: Limbic system adrenocortical dysfunction. Arch Gen Psychiatry 33:1039-1044. Carroll BJ, Curtis GC, Mendels J (1976b): Neuroendocrine regulation in depression II: Discrimination of depressed from nondepressed patients. Arch Gen Psychiatry 33:1051-1010. Carroll BJ, Feinberg M, Greden JF, et al. (1980a): Diagnosis of endogenous depression: Comparison of clinical, research and neuroendocrine criteria. J Affect Dis 2:177-194. Carroll BJ, Feinberg M, Greden JF, et al. (1981): A specific laboratory test for the diagnosis of melancholia. Arch Gen Psychiatry 38:15-22. Carroll BJ, Sebroeder K, Mukhopadhyay S, Greden JF, Feinberg M, Ritchie J, Tarika J (1980b): Plasma dexamethasone concentrations and cortisol suppression response in patients with endogenous depression. J Clin Endocrinol Metab 51:433--437. Dixon WJ, Massey FG (1969): Introduction to Statistical Analysis (ed 3). New York: McGrawHill, pp 277, 524. Haque N, Thrasher K, Werk EE, Jr, Knowles HC, Jr, Sholiton LI (1972): Studies on dexamethasone metabolism in man: Effect of diphenylhydantoin, l Clin Endocrinol 34:44--50. Holsboer F (1983): Prediction of clinical course by Dexamethasone Suppression Test (DST) response in depressed patients: Physiological and clinical construct validity of the DST. Pharmacopsychiatria 16:186-191. Holsboer F, Haack D, Gerken A, Vecoei P (1984): Plasma dexamethasone concentrations and different suppression response of cortisol and c~rticoste~ne in depressives and controls. Biol Psychiatry 19:281-291. Holsboer F, Wiedemann K, Boll E (1986a): Shortened dexamethasone half-life in depressed dexamethasone ~ msuppressors. Arch Gen Psyd~iatry 43:813-815. Holsboer F, Wiedemann K, Gerken A, Boll E (1986b): The plasma dexamethasone variable in depression: Test-retest studies and early biophase kinetics. Psychiatry Res 17:97-103. Johnson GF, Hunt G, Kerr K, Caterson I (,',984): Dexamethasone Suppression Test (DST) and plasma dexamethasone levels in depress~ :i patients. Psychk~r~, Res 13:305-313. Johnson GF, Hunt G, Kerr K, Caterson I ~1987): The plasma dexamethasone "window" and the Dexamethasone Suppression Test in depre;~sion. Biol Psychiatry 22:1033-1035. Kalent H, Roschlare WILE, Sellers EM (1985): Pr inciples of Medical Pharmacology (ed 4).. Toronto: University of Toronto Press. Maguire KP, Schweitzer NB, Bridge S, Tiller JWG (1987): The Dexamethasone Suppression Test: Importance of dexamethasone concentrations. Biol Psychiatry 22:957-967. Morris H, Cart V, Gilliard J, Hopper M (1986): Dexamethasone concentrations and the DST in psychiatEc disorders. Br J Psychiatry i48:66-69.
Plasma Dexamethasone and DST
BIOL PSYCHIATRY 1990;27:159- ! 73
173
Murphy BEP (1967): Some studies of the protein-binding of steroids and their application to the routine micro and ultramicro measurement of various steroids in body fluids by competitivebinding radioassay. J CLin Endocrinol Metab 2?:9?3-990. Nemeroff CB, Widerlov E, Bissette G, Walleus H, Karlsson I, Eklund K, Kilts CD, Loosen PT, Vale W (1984): Elevated concentrations of CSF corticotropin-reieasing factor-like immunoreactivity in depressed patients. Science 226:1342-1344. Poland RE, Rubin RT, Lane LA, Hart PJ, Lesser IM (1987): Serum dexamethasone concentration and hypothalamic-pituitary-adrenal cortical activity as determinants of the Dexamethasone Suppression Test response. Arch Gen Psychiatry 44:?90-795. Ritchie JC, Carroll BJ, Olton PR, Shively V, Feinberg M (1985): Plasma cordsol determination for the Dexamethasone Suppression Test. Arch Gen Psychiatry 42:493-49?. Rubin RT, Poland RE, Blodgett ALN, Winston RA, Forster B, Can~ll BJ (1980): Cortisol dynamics and dexamethasone pharmacokinetics in primary ED: Preliminary timings. In Brambilla F, Racagni G, de Weid D (eds), Progress in Psychoneuroendocrinology. Amsterdam: Elsevier/ North Holland, pp 223-234. Spitzer RL, Endicott J (1977): Schedule for Affective Disorders and Schizophrenia (SADS) (ed 3). New York: New York State Psychiatric Institute. Spitzer RL, Endicott J, Robins E (1978): Research Diagnostic Criteria (RDC) for a Selected Group of Functional Disorders (ed 3). New York: New York State Psychiatric Institute. Wiedemann K, Holsboer F (1987): Plasma dexamethasone kinetics during the DST after oral and intravenous administration of the test drug. Biol Psychiatry 22:1~40-1348.
! 59
Plasma Dexamethasone Concentrations and the Dexamethasone Suppression Test James C. Ritchie, Beth M. Belkin, K. Ranga R. Krishnan, Charles B. Nemeroff, and Bernard J. Carroll
Altered bioavailability or altered pharmacokinetics of dexamethasone (dex) may contribute to a positive Dexamethasone Suppression Test (DST) in psychiatric patients. We measured plasma dex and plasma cortisol concentrations in 32 patients with primary major depressive disorder (MDD), 14 patients with other psychiatric disorders, and 16 normal controls. Cortisol was measured by the competitive protein binding (CPB) assay and dex by RIA (lgG Corp.). AdditionaUy, cortisol was measured by a fluorescent polarization immunoassay (FPIA) available on the Abbou TDx analyzer in an attempt to validate this method for use in the DST. The agreement between FPIA and CPB cortisol results was excellent. Depressed nonsuppressors, by definition, had significantly higher mean plasma cortisol concentrations than depressed suppressors, psychiatric controls~ and normal volunteers at 8:00 Ata, 3:00 eta, and 10:00 eta postdex. When DST nonsuppressors and suppressors were compared regardless of diagnostic group, plasma dex concentrations were significantly lower (p < 0.01) in the DST nonsuppressors. There was a significant negative correlation ~etween plasma cortisol levels and plasma dex levels across all subjects at 8:00 ~¢ (r = -0.365, n = 44, p < 0.05). When the subjects were sorted by diagnostic category, there was a strong, but not statistically significant, trend toward lower plasma dex concentrations in the melancholic nonsuppressors versus the melancholic suppressors and between the psychiatric control non-supppressors and the corresponding suppressor group. These relationships disappeared when we restricted our analyses to an empirically derived middle range of plasma dex concentrations within which the DST results were considered to be valid. We conclude that bioavailability or pharmacokinetics of dex may significantly contribute to DST results. Further investigation is needed to determine whether or not the quantification of dex and its metabolites and their determination at which specific timepoints during the DST will enhance the predictive or interpretive value of the DST in psychiatric patients.
Introduction Patients with endogenous depression (ED) frequently have abnormalities of the hypothalamic-pituitary-adrenal (HPA) axis that resu!t in disinhibited cortisol secretion. Findings
From the Departments of Psychiatry (J.C.R., B.M.B., K.R.R.K., C.B.N., B.J.C.) and Pharmacology (C.B.N.), Duke University, Durham, NC. Supported in part by NIMH Grants MH 39593, MH 40159, and MH 42510. Address reprint requests to Dr. J. C. Ritchie, Department of Psychiatry, Duke University, Durham, NC 27710. Received February 16, 1988; revised January 23, 1989. © 1990 Society of Biological Psychiatry
0006-3223/90/$03.50
160
BIOL PSYCHIATRY 1990;27:159-173
J.C. Ritchie et al.
in such patients include hypersecretion of cortisol, inappropriate nocturnal secretion of cortisol, and escape from suppression of cortisol during the Dexamethasone Suppression Test (DST) (Carroll et al. 1976, 1980a, 1981). Among properly screened psychiatric patients, this last finding has been suggested to be specific for the melancholia (Carroll et al. 1981). It is believed to be due to a functional disturbance of the limbic system, which results in a limbic-hypothalamic overdrive, leading ultimately to hypersecretion of corticotropin-releasing factor (CRF), corticotropia (ACTH), and overproduction of cortisol. Elevated CRF concentrations have, in fact, been demonstrated in the cerebrospinal fluid (CSF) of depressed patients (Nemeroff et al. 1984; Banki et al. 1987), though it is not certain that the CSF concentration of CRF reflects the activity of the HPA axis. This hypothesis implies that there is a central nervous system (CNS) dysfunction in depression and rests on the assumptions that (1) cortisol secretion in both DST suppressors and nonsuppressors parallels that of ACTH, and (2) peripheral abnormalities, such as altered absorption or metabolism of dexamethasone, do not account for the abnormal DST results. Studies of plasma dexamethasone (dex) concentrations in DST suppressors versus DST nonsuppressors have yielded inconsistent findings. There have been several reports of significant negative correlations between postdex plasma cortisol concentrations and plasma dex concentrations (Arana et al. 1984; Holsboer et al. 1983, 1984, 1986; Morris et al. 1986; Johnson et al. 1987). In contrast, a number of investigators have been unable to find an association between plasma dexamethasone concentrations and DST response. For example, Carroll et al. (1980) found mean 4:00 PM dexamethasone concentrations of 0.75 ng/ml in 19 abnormal tests and 0.72 ng/ml in 26 normal tests (using a plasma cortisol criterion of 6 Ixg/dl). Low dexamethasone concentrations (<1.0 ng/ml) were found in 14 of the 19 abnormal tests (74%) and in 20 of the 26 normal test (77%). Rubin et al. (1980) similarly found no significant difference in a smaller early study. More recently, Poland et al. (1987) reported no significant differences in plasma dexamethasone cencentration between depressed patients with and without abnormal DST results. On the other hand, control subjects with abnormal DST results did have significantly lower plasma ,texamethasone concentrations than those with normal suppression. Discrepancies in the literature may be due to methodological factors, such as subject selection, sampling times, differences in assay sensitivity and specificity, and possible artifacts associated with uncharacterized dex metabolites and the use of unextracted plasma in the radioimmunoassays (RIAs) for plasma dexamethasone. The BST for use in psychiatry was standardized using a modified version of .k.,~.,~ competitive protein binding method for plasma cortisol (Murphy 1967). This method, developed in the late 1960s, measures total glucocorticoids and is particularly suited to accurately measure the relatively low levels encountered in the DST. The cortisol method used in most clinical laboratories, however is RIA. The disparities in these methods and the resultant use of different cortisol concentrations as the positive/negative DST cutoff point have contributed greatly to the contr,,, :rsy surrounding the DST. Previously (Ritehie eta!. 1985), we evaluated several RIA methods of plasma cortisol determination for use in the DST. Recently, a nonisotopic immunoassay for cortisol has been introduced. This method, available on the Abbott TDx, utilizes fluorescent polarization immunoassay (FPIA) technology. It is highly specific and sensitive for cortisol. Additionally, the method requires no pretreatment of the sample, is automated, and does not use radioactive isotopes (eliminating health hazards and waste disposal concerns). The present study thus was undertaken to examine further the relationship of plasma dex (utilizing the only commercial assay method available) to plasma cortisol (utilizing
Plasma Dexamethasone and DST
BIOL PSYCHIATRY 1990;27:159-173
161
the assay methodology originally employed to standardize the DST) in a rigidly defined group of depressed inpatients, nondepressed psychia~c inpatients, and normals. Aaoitionally, we sought to validate a new fluorescent polarization immunoassay of plasma cortisol for use in the DST. Methods
Subjects We studied 32 patients with major depressive disorder (MDD), primary endogenous subtype, who were hospitalized on the Affective Disorders Unit, Depart,merit of Psychiatry, Duke University Medical Center (DUMC). The diagnosis of MDD was established on the basis of a structured interview, the Schedule for Affective Disorders and Schizophrenia (SADS) (Spitzer and Endicott 1977), by a member of the research team; unstructured clinical interviews by a senior psychiatrist; review of past psychiatric history; and a comprehensive medical and clinical chemistry assessment to screen for medical disorders. Following this evaluation, patients were assigned a consensus R~search Diagnostic Criteria (RDC) (Spitzer et al. 1978) diagnosis and a (clinical) DSM-III d:..gnosis. Patients were included only when there was agreement among the staff about the diagnosis. We also studied 14 patients with other p~ychiat~ic disorders. This group included schizophrenic patients at John Umstead Hospital, Burner NC (n = 11) and patients being evaluated for chronic pain on the Clinical Specialities Unit (CSU) at DUMC (n = 3). To avoid any potentially confounding effects of the acute stress of hospitalization, the DST was never performed prior to the fourth day after admission. Patients were excluded if they met the medical, drug, or technical exclusion criteria described by Carroll et al. (1981). Finally, 16 normal volunteers, screened for medical disorders and psychiatric illness and without a family history of psychiatric illness, were studied. All subjects received an oral dose of 1.0 mg dexamethasone at 11:00 PM, and blood samples were obtained from inpatients on the following day at 8:00 AM, 3:00 PM, and 10:00 PM. Only the 8:00 AM and 3:00 PM blood samples were obtained from the normal controls. Nonsuppressor (NS) status was defined by any postdex plasma cortisol concentration > 5 ~g/dl (Carroll et al. 1981).
Assays Blood was collected in hepannized tubes for measurement of cortisol and dex. The plasma was separated from whole blood within 15 min, aliquoted in 0.5-ml portions, and frozen at -80°C until the analyses were performed. Cortisol was measured by a modified competitive protein binding (CPB) assay (Murphy 1967). It is well established that dex does not bind to plasma corticosteroid binding globulin (CBG: transcortin), and therefore, dex does not interfere with the CPB assay for cortisol. This procedure has been used in our laboratory for over 10 years, and coefficients of variation (CV) have been 12.6% (interassay) and 6.4% (intraassay) for a cortisol-poor sample (2.6 gg/dl) and 8.9% (interassay) and 5.6% (intraassay) for a cortisol-rich sample (10.15 ~g/dl) over the last 108 assays. We evaluated the Abbott TDx cortisol assay using the same paradigm we used previously to compare RIA methods for plasma cortisgl (Ritchie et al. 1985). We used the same five postdexamethasone pools as in our previous study. The pools were analyzed
162
BIOL PSYCHIATRY 1990;27:159-173
J.C. Ritchie et al.
in replicates of 10 on the TDx utilizing exactly the procedure supplied by the manufacturer. The pools were also analyzed in 19 replicate competitive protein binding assays. The results from the two procedures were then compared and a tentative cutoff value for the DST, utilizing the TDx methodology, was derived. To further validate the Abbott FPIA method and our tentative DST cutoff, we determined the plasma cortisol concentrations on the samples obtained from the patients and normal volunteers described above. Using these results, we then compared the cutoff points for a positive DST in the two procedures. The antiserum (IgG-dexamethasone-1) and protocol for the dex assay were supplied by lgG Corporation (Nashville, TN). The procedure was used with one modification. In an effort to improve the sensitivity of the assay, we used 1,2,4,6,7-3H-dexamethasone (New England Nuclear, Boston, MA) as the tracer, rather than the two prime dexamethasone trace specified by IgG Corp. The assay buffer was 0.063 M sodium dibasic p~,osphate, 0.013 M EDTA, 0.02% sodium azide. This was adjusted to pH 7.4 with HC1, and then 1 mg/ml of bovine ~amma globulin and 8-anilino-l-napthalenesulfonic acid (magnesium salt) were added to decrease nonspecific binding. The primary antiserum for the assay was prepared in rabbits against a 3-oxime conjugate of dexamethasone. For use in the assay, it was diluted 2500-fold in assay buffer plus 1% (v/v) normal rabbit serum. Standards were prepared fresh for each run in assay buffer. The standard curve was run from 0.025 ng/ml to 10 ng/ml, in triplicate. The assay is run on an unextracted plasma. For assay, 0.1 ml of standard or plasma is incubated with 0.1 ml of diluted primary antiserum and 0.1 ml of 3H-dex tracer (approximately 5500 dpm) for 20 hr at 4°C. Twenty-five microliters of a second antibody, purified goat antirabbit gamma globulin (Arnel Products, NY) is then added, mixed, and incubated for 2 hr at 4°C. Separation buffer (0.063 M sodium phosphate, 0.013 M EDTA, 0.02% sodium azide, 2.5% bovine serum albumin, pH 7.4), is added to each tube (1.6 ml). The tubes are centrifuged at 6000 x g and 4°C for 30 min. The supernatants are decanted and allowed to drain for 3 min. Then, 0.4 ml of NCS tissue solubilizer (Amersham Corp., Arlington Heights, IL) and 50 I~1 of 25% acetic acid are then added to each tube. Scintillation cocktail (2 ml) is added to each tube. The tubes are shaken and then counted on a model 460 Packard Instr. scintillation counter (Downer's Grove, IL). The scintillation counter automatically corrects for quench and luminescence. The resulting data are then reduced using standard RIA log/logit functions. The minimal detectable quantity of dex in the assay is 0.02 ng/ml. Over the last 30 assays, the 80%, 50%, and 20% binding points have averaged 0.07, 0.33, and 1.46 ng/ml, respectively. The interassay CV is 8% at a level of 0.52 ng/ml, and the intraassay CV is 4% at the same level. Quality control pools, prepared from dex and prednisolone-suppressed normals, are run in triplicate in every assay. Cross-reactivities with other major steroids, as stated by the manufacturer and verified in our laboratory, are all less than 1%. The cross-reactivities of dex metabolites are unknown.
Statistics Both the postdex plasma cortisol concentrations (determined by the CPB methodology) and the plasma dex concentrations were log-transformed because of inhomogeneity of variance. Mean plasma cortisol and dex concentrations of MDD suppressors and MDD nonsuppressors were compared utilizing the two-tailed Student's t-test for unmatched samples. Means among the four groups (MDD suppressors, MDD nonsuppressors, psychiatrically ill others, and normal controls) were compared utilizing one-way Analysis
Plasma De~methasone and DST
BIOL PSYCHIATRY 1990;27:159- !73
163
Table 1. Plasma Cortisol and Dexamethasone Concentrations during the DST Irrespective of Subject Groupsa Suppressors
Nonsuppressors
Time
Mean cortisol (ttg/dl)
Cortisol SD range
Mean dex (ttg/ml)
Dex SD range
Mean cortisol (~g/dl)
Cortisol SD range
Mean dex (ttg/ml)
Dex SD range
8:00 AM 3"00 PM 10:00 PM
1.06 1.14 1.17
0.5--2.1 0.6--2.2 0.5--2.5
2.25 0.91 0.40
1.19-4.24 0.4-1.8 0.2--1.04
6.78 6.22 6.97
2.25-18.5 2.0-19.0 2.9-16.6
1.24 0.54 0.19
0.8-2.0 0.2-1.2 0.06-0.6
aAll data are log-transformed. The p-vah~es for the cortisol comparisons were all <0.001. For dexamethasone, the p-values were <0.005~ <0.005, and <0.01 for the respective timepoints.
of Variance (ANOVA), followed by a multiple comparisons test, either the Newman Keuls pairwise procedure or the Scheff6 procedure. Relationships among the variables were evaluated utilizing Pearson correlations. The terminal (beta-phase) half-life of dex for each group was calculated using standard first-order kinetics (Kalent et al. 1985). The power of the analysis was determined by the method of Dixon and Massey (1969) for two-sided Analysis of Variance, utilizing the power curves in the same work (p 524). The two cortisol assays were evaluated utilizing least-squares linear regression and Pearson correlations. Goodness of fit was assessed by analysis of the residuals of the regression and t-test. Results The average age of the subjects was 45.4 _+ 16.7 years. There were no significant differences in age among the groups. The depressed group had a mean age of 52.9 _+ 15.5 years, the psychiatric controls had a mean age of 36.5 _+ 14.1 years, and the normal controls had a mean age of 34.3 _+ 10.98 years. There were no significant correlations between age and plasma alex or postdex cortisol concentrations. There were 36 men and 26 women overall in our population. In the depressed group, there were 19 men (9 NS) and 13 women (9 NS). The psychiatric controls were composed of 10 men (5 NS) and 4 women (0 NS). The normal control group consisted of 7 men (0 NS) and 9 (0 NS). There were no significant correlations between sex and plasma dex or postdex plasma cortisol concentrations. Within the total group of patients tested, 23 were nonsuppressors and 21 were suppressors. Of the 32 patients with MDD, 18 (56%) failed to suppress, whereas 5 (36%) of the 14 patients with other psychiatric disorders were nonsuppressors. None of the normal volunteers escaped dex suppression. The means of the log-transformed plasma cortisol and dex concentrations at each timepoint of the DST for the nonsuppressors and suppressors irrespective of diagnosis are presented in Table I. As expected, the mean plasma cortisol levels for the nonsup. pressors were significantly higher at each time of sampling than those of the suppressor~,i. In contrast, the mean plasma dex levels of the nonsuppressors were significantly lower at each time than those of the suppressors. When subjects were examined with respect to diagnostic status, depressed non~uppressors and nonsuppressing psychiatric controls again had significantly higher mean (logtransformed) plasma cortisol concentrations (p < O.Ol) than depressed suppressors, psy-
164
J.C. Ritchie et al.
BIOL PSYCHIATRY 1990;27:159-173
Table 2. Mean Plasma Cortisol and Dexamethasone Concentrations during the DST by Subject Groups° Nonsuppressors
MDD (n = 18)
Psychiatric controls (n = 5)
8:00 AM 3:00 PM 10100 PM
3.34 9.79 10.17
4.43 6.87 9.19
8:00 AM 3:00 PM 10:00 PM
1.40 0.77 0.31
1.34 0.56 0.24
Suppressors
Normals (n = 0)
MDD (n = 14)
Psychiatric controls (n = 9)
Normals (n = 16)
1.35 1.01 1.20
1.24 1.41 1.28
2.05 0.81 0.48
3.20 I. 14 0.39
Plasma cortisol (p.g/ml) -~ --
1.32 1.64 1.88
Plasma dexamethasone (izglml) ----
2.69 1.50 0.69
aAll data are log-transformed.
chiatric control suppressors, and normal volunteers at all three timepoints. There was a strong, but not statistically significant trend, toward lower mean plasma dex concentrations in MDD-NS versus MDD-S and between psychiatric control-NS versus psychiatric control-S subjects at all sampling times. These data are shown in Table 2. Figures 1, 2, and 3 are scattergrams of the log-transformed cortisol values plotted against the log-transformed dex values for each timepoint of the DST. The individual data points are coded by subject category. Suppression status is in regard to the overall DST and not necessarily the individual timepoint displayed in each figure. There was a significant inverse relationship between the plasma cortisol levels and plasma dex levels across all subjects irrespective of diagnosis at 8:00 AM (r = - - 0 . 3 6 0 ,
Figure 1. Log cortisol vs log dexamethasone concentrations at 8:00 AMtimepoint of DST.
Pldsma Dexamethasone and DST
BIOL PSYCHIATRY 1990;27:159-173
165
Figure 2. Log cortisol vs. log dexamethasone concentrations at the 3:00 PM timepoint of DST. n - 45, p < 0.025), 3:00 PM (r -- --0.380, n = 60, p < 0.005), and 11:00 PM (r = --0.400, n = 55, p < 0.005). However, tiffs relationship was not maintained when the groups were sorted by diagnoses, with the exception of the psychiatric controls. In this group, a significant negative correlation between plasma control cortisol and dex values was observed, but only at the 8:00 AM timepoint (r - - 0 . 6 3 3 , n - 11, p < 0.05).
Figure 3. Log cortisol vs. log dexamethasone concentrations at the 10:00 PM timepoint of DST.
166
J.C. Ritchie et al.
BIOL PSYCHIATRY 1990;27:i59-i73
Table 3. Plasma Dex Concentrations and the Dex Validity Windows Nonsuppressors
Suppressors
Time
Dex window (l~g/ml)
n
Mean
n
Mean
8"00 AM 3:00 PM 10:00 PM
0.96-2.31 0.34-1.67 0.10-0.91
8 12 15
1.60 0.93 0.33
15 31 27
1.47 0.93 0.38
If the analysis of these data are restricted to only those samples with midrange dex concentrations (established empirically by inspection of the distributions) at each time point (8:00 AM, 0.96-2.31 ng/ml; 3:00 PM, 0.34-1.67 ng/ml; 10:00 PM, 0.1--0.91 ng/ml), the significant difference in plasma dex concentrations between suppressors and nonsuppressors, irrespectiveof diagnosis, is lost. These data are shown in Table 3. Additionally, the observed inverse association between plasma cortisol concentrations and plasma dex concentrations at each timepoint across all subjects, irrespective of diagnosis, disappears (p > 0.1 for all the correlations). We used the plasma dex concentrations obtained in this study to calculate the elimination halfAife (t~]2) for dex in each group. These calculations assume first-order kinetics and appear to be justified based on the position of the three DST samples late in the terminal elimination phase of dex. The mean t 1/2 of the normal controls was 4.61 hr. Overall, in the patient groups, both nonsuppressors (t ~/2 5.62 hr) and suppressors (t !12 6.11 hr) had half-lives greater than the normals. When broken down by patient category, this increase in t ~/z persisted in all groups. In the depressed patients, nonsuppressors had a t ~/2of 6.43 hr, whereas the suppressors had a t ~]2of 7.13 hr. In the psychiatric controls, the nonsuppressors had a t ~2 of 5.64 hr and suppressors 6.68 hr. None of these comparisons reached significanc¢~, due to the large variability exhibited by the subjects. However, we aid note a trentl (p = 0.06) for larger t~/2s in the patient suppressor group when compared to the normal controls (6.90 hr versus 4.61 hr). The results for the comparison of the two eortisol assay methods utilizing the postdexamethasone plasma pools were excellent (r = 0.9985, slope = 0.88, y-intercept = 0.90). The results for the fifth pool were not used, as this pool was beyond the range of linearity for the CPB procedure, and thus, direct comparisons could not be made. The 95% confidence interval for a TDx value, which corresponded to a CPB value of 5.0, was 4,88-5.72 p,g/dl. Based on these results, we tentatively set the cutoff value for the DST utilizing TDx cortisols at 5.0 I~g/dl. We then determined the plasma cortisol concentrations on the 169 postdexamethasone samples generated in this study, in duplicate, on the TDx and in the CPB assaY. These results are shown in Figure 4. The procedures were highly correlated, and when mean TDx values were compared to mean CPB values, they yielded a correlation coefficient of 0.9942 (slope = 0.89, y-intercept = -0.08). These results show a significant linear relationship between the two procedures, with a p-value of < 0.01. As more than half of the values in the above comparisons were <5 I~g/dl, we also plotted the residuals from the regression analysis. Though not shown, this plot showed no consistent error trend along the regression line. The determination of accurate levels of cortisol between 3 and 8 I~g/dl is critical for the DST. Figure 5 presents such a subset of the data from Figure 4. In Figure 5, we replotted those samples that had CPB cortisol values >3 and <8 ~g/dl. As can be seen, the two procedures are still highly correlated, with the exception of one obvious outlier that had a CPB value of 8 and a TDx value of 2 ~g/dl.
Plasma Dexamethasone and DST
BIOL PSYCHIATRY 1990;27:159-173
!~_.POSI-II~ R~tA Stm
,° I 35
r -- 0 . 9 9 4 2
-
SLOPE = 0 . 8 9 Ycept = - 0 . 0 8
30
n = 169 25
a
..I
0 ¢n 2 0 I-., I]=: 0 15 ¢J, o I-
q a
o
o
•
Q
10
aa
5
oJ
P"
13
uO/dL
I
I
O
5
.
I
.
I
10
,
15
I
.
I
20
,
I
25
.
30
I
,
35
I._
40
C.P.B.L. CORTISOL CPB Figure 4. TDx cortisol vs. CPB cortisol assay; 169 postdex plasma samples.
PLASMA SAXPLES > 3 a n d
<8
ug/dL
10 r =
0.7376
SLOPE = Ycept =
0.73 -0.14
n=31 -J 0 fJ) I0=: 0 ¢J ,ell
0
°a ~
a
N
a aaaaa
I-
I
0
I
I
2
,
_
I _.
4
,
I
.
6
I
.I
ug/dL
8
G.P,B.L. CORTISOL CPB Figure 5. TDx cortisol vs. CPB cortisol assay; plasma samples >3 and <8 I~g/dl.
I..
10
167
168
J.C. Ritchie et al.
BIOL PSYCHIATRY 1990;27:159-173
Table 4. Comparison of CPB and TDx Cortisol Values in the DST: Effect of Varying TDx Cutoff on Test Outcome TDx cutoff point 3.0 3.5 4.0 4.5 5.0 5.5 6.0
Izg/dl Izg/dl Ixg/dl p.g/dl p,g/dl p,g/dl ~g/dl
CPB + TDx +
CPB TDx +
CPB + TDx -
43 43 41 40 36 34 29
8 4 2 1 1 1 0
1 1 3 4 8 10 15
CPB TDx 117 121 123 124 124 124 125
Percent overall agreement° 94.67 97.04 97.04 97.04 94.67 93.49 91.12
°Using the CPB assay and a DST cutoff of 5.0 Izg/dl, there were 44 positive and 125 negative results.
Discussion The purpose of the comparison of the two cortisol assay methodologies in this study was to determine a DST cutoff, using the TDx assay, that best compared to the standardized DST cutoff, using the CPB assay. We made an initial evaluation of the two methods utilizing the postdex cortisol pools that we have used previously (Ritchie et al. 1985) i~,r this purpose. This evaluation showed the methods to be comparable, and we determined that a TDx DST cutoff of 5 ~g/dl was ~ost appropriate. We next ran parallel studies on the 169 postdex plasma samples generated in this study. This was done to further optimize the DST cutoff in the TDx assay by using real patient samples. These results are summarized in Table 4. As can be seen from the table, all of the possible TDx cu'~erf points, from 3 to 6 ~g/dl, gave an overall agreement of >90% with the CPB cutoff of 5 ~g/dl. We therefore recommend that a value of 5 Izg/dl be used as the cutoff criteria for the DST when using either the TDx or CPB method of cortisol analysis. Also, as per our previous recommendation, we continue to believe that all postdexamethasone cortisol concentrations >3 and <7 ~g/dl should be confirmed by reanalysis of the samples. This is due to the importance of a positive test result to the psychiatrist and does not reflect any inaccuracy in either measurement technique. In this study, we have attempted to define the relationship of postdex plasma cortisol concentrations to circulating dexamethasone concentrations during the l-mg DST. We used the cortisol assay methodology originally used to standardize the DST and the only commercially available plasma dex RIA in an attempt to establish this relationship (plasma dex to plasma cortisol) in the most reliable and easily available methodologies. Our results demonstrate considerable individual variability in plasma dex concentrations following a 1-mg oral dose in depressed patients, psychiatrically ill others, and normal controls. These findings are consistent with those of previous investigations. In fact, a 100-fold variation in 4:00 PM plasma dexamethasone concentrations (range 0.1-9.1 ng/ml) can be seen in the combined results of patients and control subjects reported to date (Arana et al. 1984; Carroll et al. 1980; Holsboer et al. 1984, Johnson et al. 1984; Morris et al. 1986; Poland et al. 1987). Most l:revious studies have attributed this broad range to significant individual variations in dex metabolism. An alternative hypothesis, however, is that this broad range may be due to assay interference by unknown dex metabolites. Further investigation will be needed to clarify this point. When we examined the entire range of plasma dex concentrations in all subjects irrespective of diagnosis, we found a significant inverse relationship between dex levels
Plasma Dexamethasone and DST
BIOL PSYCHIATRY 1990;27:159-173
169
and plasma cortisol concentrations. However, we found no significant difference in plasma dex concentrations between DST suppressors and nonsuppressors when only patients with diagnosis of MDD were considered, though a strong trend in this direction was evident (Table 2). Post hoc power analysis (Dixon andMassey 1969) yielded power factors of >0.7, >0.6, and >0.6 at the three respective timepoints of the DST for the comparison of plasma dex concentrations between MDD-suppressors and MDD-nonsuppressors (based on an oLof 0.05 and 14 individuals per group). The analysis also revealed that given the large variance displayed by each group, we would have to increase the size of each group by a factor of 5 to reach a power factor of 0.8. These results are consistent with those of several studies that found a negative correlation between plasma cortisol and plasma dex levels in psychiatric patients and normal controls. Morris et al. (1986) also observed wide individual variation in serum rex concentrations following a !-mg dose in 51 patients representing a cross-section of all psychiatric disorders. They, too, found a significant inverse relationship between serum dex levels and serum cortisol levels. Johnson et al. (1984) observed considerable individual variation in plasma dex levels across patients and normal subjects and noted an inverse relationship between serum dex and cortisol levels. Studies of MDD suppressors versus MDD nonsuppressors are conflicting as to differences in plasma dex levels. In the first study of plasma dex concentrations during the DST in psychiatric patients, Carroll et al. (1980b) measured the plasma dex levels in 4 endogenously depressed patients before treatment (when abnormal DST results were observed) and again after recovery (when the DST results had normalized). No differences in plasma rex concentrations or t~/2s were observed at any timepoint. They also compared plasma dex concentrations in 10 endogenous depressed, age- and gender-matched subjects with 10 nonendogenous depressed outpatients. All of the former had abnormal DST results, and all of the latter suppressed normally. No differences in plasma dexamethasone concentrations were found. Also, in 1980, Rubin et al. (1980) studied 15 patients with MDD and 8 normal controls matched to the patients on age, gender, race, and (for women) endocrine status. They found no difference in 7:00 AM postdex serum dex ievels or in average dex half-lives among suppressors, nonsuppressors, and normal controls. In a recent study, Poland et al. (1987) found no significant dexamethasone concentration differences in MDD patients (15 DST nonsuppressors and 25 suppressors). Although they observed a trend toward reduced serum dex concentrations and an inverse, correlation between postdex serum cortisol and dex levels in the nonsuppressing patients, nei,'her of these effects reached statistical significance. The first finding of plasma dexamethasone concentration difference between MDD suppressors and MDD nonsuppressors was reported by Holsboer (1983). Another group reported similar results in later studies (Johnson et al, 1984, 1987). In another recent report, Maguire et al. (1987) described a bimodal distribution of dexamethasone concentrations among MDD nonsuppressors: one subgroup had low levels similar to those of volunteer subjects with abnormal DST results, whereas the second subgroup of MDD nonsuppressors had dexamethasone levels equivalent to those of normally suppressing volunteer subjects. The original finding of a significant difference in dex concentrations within MDD patients (i.e., MDD suppressors versus MDD nonsuppressors) was demonstrated again by Holsboer et al. (1984). In this study, these investigators found significant concentration differences in 16 MDD nonsuppressors when compared with 6 MDD suppressors and 6 normal controls at one timepoint during the DST. In a more recent study (Holsboer et
170
BIOL PSYCHIATRY 1990:27:159-173
J.C. Ritchie et al.
al. 1986), this group used the test-retest paradigm to examine the plasma dex concentrations of 31 MDD patients whose DST results converted with treatment. They found significantly lower mean plasma dex concentrations during illness than after recovery. Additionally, using frequent catheter sampling, they observed a decreased t ~/2 (160 min) in 5 depressed patients with abnormal DSTs, compared with 6 other depressed patients whose DSTs were normal (422 min). In 14 other patients with MDD who exhibited adequate DST suppression during the depressive state and after recovery, the dex concentrations were unchanged. Our results indicated a trend toward a significant plasma dexamethasone difference between MDD suppressors and MDD nonsuppressors. That this trend was not statistically significant may be due to the small sample size. It may also be explained by differences in the dex assay methodologies used. For example, in 4 separate reports, the 4:00 PM mean plasma dexamethasone concentration associated with abnormal DST results ranged from 0.50 to 2.18 ng/ml (Arana et al. 1984; Holsboer et al. 1984; Johnson et al. 1985, this study, Table 2). This variation of 2-4-fold across studies strongly suggests that the dexamethasone assay methods reported from different laboratories are not equivalent. It could be that dexamethasone metabolites (so far uncharacterized) may cross-react in some assays, falsely elevating the levels in some patients. This would also explain our inability to show a difference in the half-lives between suppressors and nonsuppressors, as the German group did. Interestingly, when compared to normals (4.61 hr), all of our ho~ pitalized subjects (depressed 6.71 and other psychiatric patients 6.35) exhibited a somewhat lengthened elimination half-life. The Holsboer group did not publish any data on normals in their 1986 study (Holsboer et al. 1986). Previously reported dex elimination half-lives have varied in normals from 2.78 to 6.13 hr in a specifically designed pharmacokinetic study (Poland et al. 1987). Thus, a high level of interindividual variation seems to exist in the metabolic processing of dex, even in normal individuals. On inspection of the wide range of plasma dex concentrations in our sample, we identified a middle range of dex values below which subjects tended to escape from DST suppression and above which they never failed to suppress. We propose that those extreme values of the dex distribution represent areas of increased likelihood of false-positive and false-negative results, respectively. When we restricted our analysis to this empirically identified middle range of dex levels, we found no difference in mean plasma dex concentrations between suppressors and nonsuppressors, irrespective of diagnosis. Similarly, the previously observed inverse association between plasma cortisol and dex levels also disappeared. These findings support the hypothesis that there is a range of plasma dex concentrations within which the level of exogenous steroid attained is sufficient to suppress cortisol production in patients with normal HPA axis function, but which is not high enough to cause suppression in all patients with altered HPA axis function. We term this range the dex "window" and propose that the DST may not be valid in patients whose plasma dex values fall outside of this window. Table 3 presents our first approximation of these dex windows based on the data generated in this study. The table shows the mean plasma dex levels within these windows for suppressors and nonsuppressors, including all diagnoses. As c~n be seen, the significant difference initially observed between suppressors and nonsuppressors has disappeared. Moreover, within these windows, there was no significant correlation between plasma cortisol and dex concentrations either across all subjects or within any of tile groups at any of the three timepoints. When these validity windows are applied to our study population, 9 of 14 psychiatric controls are eliminated (4 NS and 5S), reducing the rate of nonsuppression from 36% to 20% in this group. In
Plasma'Dexamethasone and DST
BIOL PSYCHIATRY 1990;27:159-173
171
the depressed group, 11 patients are eliminated (6 NS and 5 S), and the rate of nonsuppression remains relatively unchanged at 57%. Johnson et al. (1987) were the first to propose a window concept for valid dexamethasone suppression testing in psychiatry. They studied a sample of 9 MDD, 8 psychiatric controls, and 9 normals using a 0.5-mg and 1.0-mg repeat DST paradigm and a different dex RIA than ours. In describing their results, they proposed a dex window of 1.964.32 ng/ml at 8:00 AM and 0.78-1.57 ng/ml at 4:00 PM. It is unclear whether the differences in the windows proposed by Johnson and those derived in this study (Table 3) are due to subject differences or variations in assay methodologies. Johnson et al. also noted that within the window, "significant differences in plasma dexamethasone concentrations between suppressors and nonst:ppressors are no longer evident," a finding which our data confirm. The definitive determination of window limits will require additional studies with a much larger sample size and a standardized dexamethasone assay methodology. Future studies should seek to determine the relative contribution, if any, of age, gender, weight, and hospitalization on the metabolism of dex. Additionally, studies should be pursued that seek to elucidate more fully the metabolites of dex, their cross-reactivity in the currently available assay methodologies, their bioactivity in suppressing the HPA axis, and whether or not the hypercortisolemia, which occurs in melancholia patients, induces a more rapid clearance of dex. In this regard, the recent study by Wiedemann and Holsboer (1987) is of great interest. Their data (Table 3, p 1345) strongly suggest that dexamethasone metabolites cress-react in the RIA procedure for plasma "dexamethasone," as the apparent dexamethasone concentrations were substantially higher from 2:00 AM through 11:00 PM in normally suppressing patients after oral dex administration than after intravenous dexamethasone, despite the fact that identical peak dexamethasone levels were achieved by both routes 1 hr after drug administration. In the same study, other data would be consistent with the possibility that nonsuppressing patients have accelerated dexamethasone clearance, possibly with a different metabolic profile than suppressing patients, especially after oral dexamethasone administration. Resolution of these discrepancies awaits the complete characterization of dex metabolism in normals and depressives. This in turn awaits the development of assay methodologies capable of accurately measuring parent dex and its metabolites during the standard DST. Finally, we emphasize that our plasma dex data refer only to the time period 9-23 hr after dex administration. Holsboer et al. (1986b) have shown that plasma dex concentrations during this time interval are not strongly related to the concentrations earlier in the night, 0-4 hr after dex administration, when the steroid is believed to exert its major pharmacodynamic action in suppressing the "circadian program" of HPA activity (see Carroll et al. 1976a). Additional studies, therefore, will be required to determine the optimal sampling times for relating dex pharmacokinetics to HPA suppression. Until this question is resolved, however, our preselit results, together with those of Johnson et al. (1987), suggest that the concurrent measurement of plasma dex and plasma cortisol at the standard DST sampling times (Carroll et al. 1981) can improve the clinical information provided by this test. We gratefully acknowledge the efforts of Mrs. A.C. Laws and Mrs. T. Page in the preparation of this manuscript. We also wish to thank Ms. E. Kegelmeyer for her expert technical assistance.
172
BIOLPSYCHIATRY
J.C. Ritchie et al.
1990;27:159-173
References American Psychiatric As~ ~ciation (1980): DSM-III: Diagnostic and Statistical Manual of Mental Disorders (ed 3). Wa~ingten, DC: American Psychiatric Association. Arana GW, Workman RJ, Baldessarini RJ (1984): Association between low plasma levels of dexamethasone and elevated levels of cortisol in psychiatric patients given dexamethasone. Am J Psychiatry 141:1619-1620. Banki CM, Bissette G, Arato M, O'Connor L, Nemeroff CB (1987): CSF corticotropin-releasing factor-like immunoreactivity in depression and schizophrenia. Am J Psychiatry 144:873-877. Can" B, Morris H, GiUiard L (1986): The effect of serum dexamethasone concentrations on the Dexamethasone Suppression Test. Biol Psychiatry 21:735-743. Carroll BJ (1983): The Dexamethasone Suppression Test for depression. In Habig RL (ed), The Brain, Biochemistry, and Behavior. American Association for Clinical Chemistry Publications, pp 69-84. Carroll BJ (1984): The Dexamethasone Suppression Test. In Hall RCW, Beresford TP (eds), Handbook of Psychiatric Diagnostic Procedures. New York: Spectrum Publications, pp 3-18. Carroll BJ, Curtis GC, Mendels J (1976a): Neuroendoerine regulation in depression I: Limbic system adrenocortical dysfunction. Arch Gen Psychiatry 33:1039-1044. Carroll BJ, Curtis GC, Mendels J (1976b): Neuroendocrine regulation in depression II: Discrimination of depressed from nondepressed patients. Arch Gen Psychiatry 33:1051-1010. Carroll BJ, Feinberg M, Greden JF, et al. (1980a): Diagnosis of endogenous depression: Comparison of clinical, research and neuroendocrine criteria. J Affect Dis 2:177-194. Carroll BJ, Feinberg M, Greden JF, et al. (1981): A specific laboratory test for the diagnosis of melancholia. Arch Gen Psychiatry 38:15-22. Carroll BJ, Sebroeder K, Mukhopadhyay S, Greden JF, Feinberg M, Ritchie J, Tarika J (1980b): Plasma dexamethasone concentrations and cortisol suppression response in patients with endogenous depression. J Clin Endocrinol Metab 51:433--437. Dixon WJ, Massey FG (1969): Introduction to Statistical Analysis (ed 3). New York: McGrawHill, pp 277, 524. Haque N, Thrasher K, Werk EE, Jr, Knowles HC, Jr, Sholiton LI (1972): Studies on dexamethasone metabolism in man: Effect of diphenylhydantoin, l Clin Endocrinol 34:44--50. Holsboer F (1983): Prediction of clinical course by Dexamethasone Suppression Test (DST) response in depressed patients: Physiological and clinical construct validity of the DST. Pharmacopsychiatria 16:186-191. Holsboer F, Haack D, Gerken A, Vecoei P (1984): Plasma dexamethasone concentrations and different suppression response of cortisol and c~rticoste~ne in depressives and controls. Biol Psychiatry 19:281-291. Holsboer F, Wiedemann K, Boll E (1986a): Shortened dexamethasone half-life in depressed dexamethasone ~ msuppressors. Arch Gen Psyd~iatry 43:813-815. Holsboer F, Wiedemann K, Gerken A, Boll E (1986b): The plasma dexamethasone variable in depression: Test-retest studies and early biophase kinetics. Psychiatry Res 17:97-103. Johnson GF, Hunt G, Kerr K, Caterson I (,',984): Dexamethasone Suppression Test (DST) and plasma dexamethasone levels in depress~ :i patients. Psychk~r~, Res 13:305-313. Johnson GF, Hunt G, Kerr K, Caterson I ~1987): The plasma dexamethasone "window" and the Dexamethasone Suppression Test in depre;~sion. Biol Psychiatry 22:1033-1035. Kalent H, Roschlare WILE, Sellers EM (1985): Pr inciples of Medical Pharmacology (ed 4).. Toronto: University of Toronto Press. Maguire KP, Schweitzer NB, Bridge S, Tiller JWG (1987): The Dexamethasone Suppression Test: Importance of dexamethasone concentrations. Biol Psychiatry 22:957-967. Morris H, Cart V, Gilliard J, Hopper M (1986): Dexamethasone concentrations and the DST in psychiatEc disorders. Br J Psychiatry i48:66-69.
Plasma Dexamethasone and DST
BIOL PSYCHIATRY 1990;27:159- ! 73
173
Murphy BEP (1967): Some studies of the protein-binding of steroids and their application to the routine micro and ultramicro measurement of various steroids in body fluids by competitivebinding radioassay. J CLin Endocrinol Metab 2?:9?3-990. Nemeroff CB, Widerlov E, Bissette G, Walleus H, Karlsson I, Eklund K, Kilts CD, Loosen PT, Vale W (1984): Elevated concentrations of CSF corticotropin-reieasing factor-like immunoreactivity in depressed patients. Science 226:1342-1344. Poland RE, Rubin RT, Lane LA, Hart PJ, Lesser IM (1987): Serum dexamethasone concentration and hypothalamic-pituitary-adrenal cortical activity as determinants of the Dexamethasone Suppression Test response. Arch Gen Psychiatry 44:?90-795. Ritchie JC, Carroll BJ, Olton PR, Shively V, Feinberg M (1985): Plasma cordsol determination for the Dexamethasone Suppression Test. Arch Gen Psychiatry 42:493-49?. Rubin RT, Poland RE, Blodgett ALN, Winston RA, Forster B, Can~ll BJ (1980): Cortisol dynamics and dexamethasone pharmacokinetics in primary ED: Preliminary timings. In Brambilla F, Racagni G, de Weid D (eds), Progress in Psychoneuroendocrinology. Amsterdam: Elsevier/ North Holland, pp 223-234. Spitzer RL, Endicott J (1977): Schedule for Affective Disorders and Schizophrenia (SADS) (ed 3). New York: New York State Psychiatric Institute. Spitzer RL, Endicott J, Robins E (1978): Research Diagnostic Criteria (RDC) for a Selected Group of Functional Disorders (ed 3). New York: New York State Psychiatric Institute. Wiedemann K, Holsboer F (1987): Plasma dexamethasone kinetics during the DST after oral and intravenous administration of the test drug. Biol Psychiatry 22:1~40-1348.