| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Department of Psychiatry (Mo.B.G., Me.B.G., S.P., S.Z., G.F.), University of Chicago, and Psychiatric Institute (Mo.B.G., G.W., M.G.), University of Illinois at Chicago in affiliation with University of Chicago, Chicago, Illinois 60637; and Departments of Statistics and Biology (R.S.P.), Case Western Reserve University, Cleveland, Ohio 44106
Address all correspondence and requests for reprints to: Dr. Morris B. Goldman, Department of Psychiatry, University of Chicago Hospitals, 5841 South Maryland, MC 3077, Chicago, Illinois 60637. E-mail: m-goldman{at}uchicago.edu.
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Objective: The objective of the study was to determine the effect of cortisol on plasma ACTH under conditions thought to be most sensitive to hippocampal influences.
Design: The design was repeated measures.
Setting: The study was conducted at an inpatient clinical research center.
Participants: Participants included eight polydipsic hyponatremic and eight polydipsic normonatremic as well as six schizophrenic patients without water imbalance. Eight healthy community volunteers matched for age and gender were also studied.
Intervention: Metyrapone (750 mg) was administered orally at 1430 and 1900 h. Beginning at 1930 h, hydrocortisone was infused over 150 min at 0.03 mg/kg·h. Blood samples and other measures were obtained at 20-min intervals from 1850 to 2320 h.
Main Outcome Measures: Plasma ACTH and cortisol were measured.
Results: ACTH levels did not decline significantly during the cortisol infusion in the polydipsic hyponatremic group. For any given level of cortisol, ACTH levels were higher in the hyponatremic group. Although levels declined after cortisol in the other three groups, the decline was greatest in patients without water imbalance.
Conclusions: The marked impairment in glucocorticoid negative feedback in polydipsic hyponatremic schizophrenic patients is consistent with hippocampal mineralocorticoid dysfunction.
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
These neuroendocrine findings in polydipsic patients could reflect hippocampal dysfunction. The hippocampus is the brain region most frequently implicated in schizophrenia (9). This structure modulates vasopressin and HPAA responses to both psychological stress (10, 11) and glucocorticoid feedback (12, 13). The anterior hippocampus, which is selectively smaller in patients with polydipsia and hyponatremia, compared with matched schizophrenic patients (Ref. 14 and manuscript submitted for publication), is the primate analog of the segment that restrains neuroendocrine responses to psychological stress in rodents (10, 11). An animal model of schizophrenia, which disrupts neurodevelopment of this segment (15), impairs glucocorticoid feedback (16) as well as the restraint of vasopressin and HPAA responses to psychological stress (17). Based on these observations, and other data suggesting a heightening of putative hippocampal influences on neuroendocrine secretion in patients with normal water balance (18), we made the following predictions: HPAA feedback will be blunted in 1) polydipsic patients with hyponatremia, compared with those with normonatremia; 2) these two groups, compared with healthy controls; and 3) these three groups, compared with patients with normal water balance (i.e. a representative sample of the majority of schizophrenic patients).
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Psychiatric subjects were grouped by a two-step procedure (6) before and after transfer to the Psychiatric Clinical Research Center at the University of Illinois at Chicago. Depending on the group to which they were initially assigned, subjects had prior, or no prior, evidence of polydipsia (unexplained urine specific gravities < 1.008 on multiple occasions) or hyponatremia (unexplained serum sodium levels < 125 mEq/liter on at least two occasions) (6, 19). Grouping was subsequently confirmed by obtaining spot urine and plasma samples three times weekly during the first 3 wk of admission. Confirmed polydipsic hyponatremic subjects had two or more morning plasma osmolalities that were less than 275 mosmol/kg and mean afternoon urine osmolality less than 300 mosmol/kg (n = 8), whereas polydipsic normonatremics had mean plasma osmolality greater than 285 mosmol/kg (none lower than 280 mosmol/kg) and mean afternoon urine osmolality less than 300 mosmol/kg (n = 8). Nonpolydipsic normonatremic subjects (i.e. normal water balance) had mean plasma osmolality greater than 285 mosmol/kg and mean afternoon urine osmolality greater than 500 mosmol/kg (n = 6; Table 1
). All had primary diagnoses of schizophrenia or schizoaffective disorder and were otherwise without major medical or neurological disorders (6). None was taking corticosteroids, lithium, thiazide diuretics, carbamazepine, or chlorpropamide. Informed witnessed written consent was obtained after the Institutional Review Boards of the University of Chicago, and the University of Illinois at Chicago approved the studies. We matched groups based on age, gender, and diagnosis.
|
). During the fifth week of admission, trained technicians administered the Positive and Negative Symptom Scale (PANSS) (20), Hamilton Depression Scale (21), and the Global Assessment of Functioning Scale (22). Psychiatric diagnosis was confirmed at discharge in a multidisciplinary conference and relied largely on results from the Structured Clinical Interview Diagnosis for the Diagnostic and Statistical Manual of Mental Disorders, fourth edition (23). Healthy controls were recruited by advertisements placed around the University community, screened as previously described (6), and studied as outpatients. Each was administered the Structured Clinical Interview for Diagnostic Statistical Manual–nonpatient version (version 2.0) (24) by a trained technician to rule out a current or past Diagnostic and Statistical Manual of Mental Disorders, fourth edition, Axis I disorder. Procedures
Hippocampal involvement in glucocorticoid-negative feedback can be inferred in vivo by assessing feedback during the diurnal nadir (25, 26). At this time of the day, intrinsic HPAA stimulatory activity is minimal, and mineralocorticoid receptors (which have much greater affinity for cortisol than glucocorticoid receptors and are concentrated in the hippocampus) have a greater impact on ACTH levels (27, 28). To enhance our ability to detect differences in feedback sensitivity and better assure infused cortisol was acting at mineralocorticoid receptors, we first lowered endogenous cortisol levels with metyrapone, which blocks the conversion of 11-deoxycortisol to cortisol (29).
Figure 1 outlines our approach, which was adapted from the literature (29). Psychiatric subjects were studied during the sixth week of their admission after completing related studies [magnetic resonance imaging scan, cold pressor/postural stimuli (5)] on wk 4 and 5, respectively. Upon admission to the Clinical Research Center, subjects were prohibited from smoking and fluid intake was limited. Two 750-mg doses of metyrapone (Metopirone; Novartis, East Hanover, NJ) were administered orally with milk at 1630 and 1900 h. Two iv lines were placed at 1800 h after a light meal. Ten milligrams of Solu-cortef (Pfizer, Kirkland, Quebec, Canada) were dissolved in 500 mg of 5% dextrose and infused at the rate of 0.03 mg cortisol per kilogram body weight per hour beginning at 1930 h and ending at 2200 h. Blood samples were obtained at 20-min intervals from 1850 to 2320 h. At each sampling interval, blood was drawn, and vital signs (Dinamap XL vital signs monitor; Critikon, Tampa, FL) along with presence of nausea or lightheadness were recorded. Subjects were supine throughout the study and not allowed to fall asleep. Healthy controls were studied as outpatients under two conditions, 1–2 wk apart. In a randomly assigned order they received an infusion of cortisol on one occasion and diluent on the other.
|
Laboratory
Collected blood samples were processed as previously described (7). Plasma cortisol and ACTH were measured by chemiluminescent automated immunoassays [cortisol, Immulite 1000; Diagnostic Products Corp., Los Angeles, CA; lower limit of detection of 0.2 µg/dl, intraassay coefficient of variation of 3.0–5.1%, interassay coefficient of variation of 4–6.4%; ACTH: Nichols Advantage, San Clemente, CA; limit of detection 1 pg; interassay coefficient of variation at 14 pg, 7.0%]. 11-Deoxycortisol was assayed using a double-antibody RIA kit (ICN Pharmaceuticals Inc., Costa Mesa, CA), and salivary cortisol was measured by ELISA (Salimeterics kit; State College, PA; limit of detection 0.05 pg/ml, intraassay coefficient of variation of 4.4%, interassay coefficient of variation of 7.6%.).
Data analysis
The data from the repeated measures design were analyzed using a linear mixed-effects regression model with normal error terms (32, 33). Unlike a standard ANOVA, this method accurately models the intersubject differences in initial ACTH levels and the intrasubject correlation of the repeated measures (33). ACTH was transformed by taking its square root to normalize the distribution of the residuals at each of the 15 time points (–40 min to 240 min at 20 min intervals). Time was centered at the midpoint of the study (+100 min). The shape of the response was decomposed into linear, quadratic, and cubic time trends to isolate differences in the rate of decline (i.e. linear trend = slope) of ACTH as well as differences in the rise and decline of the measures (i.e. quadratic trend).
The three time trends were treated as fixed effects, whereas the subject effect was treated as a random effect. To test our three hypotheses (see introductory text), we formulated the analysis of the group effects as three contrasts: 1) the polydipsic hyponatremic group, compared with the polydipsic normonatremic group; 2) these two polydipsic groups, compared with the healthy controls; and 3) these three groups, compared with the nonpolydipsic normonatremic patients. The three group contrasts and their nine interactions with the time trends were entered into the model as fixed effects. The two conditions in the healthy controls were analyzed in an analogous manner. Type I error level was set at 0.05. All effects are reported as Z scores. The relationship between ACTH and cortisol was assessed using a linear mixed-effects regression model as previously described (6).
Significant main effects are reported only when interaction terms were insignificant. Cross-sectional data were analyzed using an ANOVA with the group contrasts defined above. For exposition of the Results section, significant group contrasts are defined in a narrative form (e.g. nonpolydipsic normonatremic subjects levels declined faster than the other three groups) and the associated test statistic for the relevant time trend (e.g. linear) by group contrast interaction is listed in parentheses.
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
To characterize the normal ACTH response to cortisol better, healthy controls were studied twice: once receiving an infusion of cortisol and once receiving an infusion of saline. As expected, recognized neuroendocrine modulators (i.e. plasma osmolality, plasma glucose, plasma sodium, blood urea, mean arterial pressure, heart rate, expired air carbon monoxide, hemoglobin) before the infusion were similar in the two conditions (data not shown). Cortisol and ACTH levels at this juncture (i.e. 30 min after the second dose of metyrapone) were about half and double normal evening levels, respectively (Fig. 2, A and B). One subject noted mild transient nausea during the diluent infusion; however, there were no reports of lightheadness or nausea reported during the cortisol infusion.
|
). In contrast, levels did not change after diluent (condition by time interactions: quadratic trend Z = 10.1, P < 0.001). Eighty minutes into the cortisol infusion, ACTH levels began a linear decline (Fig. 2B), whereas they were rising in the diluent condition consistent with the peak effects of the second dose of metyrapone (condition by time interaction: linear trend Z = 5.87, P < 0.001). As expected, 11-deoxycortisol levels mirrored ACTH levels. Thus 11-deoxycortisol levels began a linear decline well into the cortisol infusion, whereas levels were rising in the diluent condition (Fig. 2C, condition by time interactions: linear trend Z = 3.62, P < 0.001). Plasma glucose declined slightly in both conditions (linear trend Z = 2.13, P = 0.032). Levels of other recognized modulators did not change. Demographic and baseline measures: all groups
The three patient groups resembled each other on demographic, behavioral, and clinical variables except for expected differences in indices of water balance (Table 1). The number of patients receiving atypical antipsychotic medication did not differ across groups. Demographic as well as baseline laboratory measures were also similar in all four groups, except for the baseline sodium measurement (which was lower in the hyponatremic subjects) and the nonfasting glucose measurement (which was lower in the healthy controls) (Table 2).
|
ANOVA under the specified group contrasts revealed that preinfusion cortisol levels (i.e. 0 min) did not differ. ACTH levels, in contrast, were lower in healthy controls than the two polydipsic patient groups, who resembled nonpolydipsic patients in this respect (Table 2). Consistent with the prestudy fluid restriction, plasma sodium did not differ across groups. Other basal measures were also similar. Plasma cortisol rose steadily during the infusion and afterward returned toward baseline (Fig. 3A). The fitted mixed-effects linear regression model revealed a slightly, but significantly, slower rise in the polydipsic hyponatremic, relative to the normonatremic, group (quadratic trend Z = 2.83, P = 0.004).
|
, linear trend Z = 5.04, P < 0.001). Also as predicted, the rate of decline in the healthy controls was greater, albeit marginally, than in the two polydipsic groups (linear trend: Z = 1.92, P = 0.054); and the rate in nonpolydipsic patients exceeded that of the other three groups (Fig. 3B, linear Z = 7.07, P < 0.001). ACTH levels declined significantly from their baseline levels at 80 min into the infusion in the nonpolydipsic patients, 100 min in healthy controls, 160 min in the polydipsic normonatremic patients and at no time point in the polydipsic hyponatremic patients. To exclude the possibility that changes in the rate of cortisol increase could account for the group differences, we calculated the change in ACTH from baseline as a function of change in lagged (–40 min) cortisol from baseline at subsequent time point as previously described (29). This analysis also demonstrated a steeper decline in ACTH in the polydipsic normonatremic than the hyponatremic group (Z= 3.97, P < 0.001) and in the nonpolydipsic patients, compared with the other three groups (Z = 4.11, P < 0.001). The response in the two polydipsic groups did not differ significantly from the healthy controls.
To examine further the relationship between the two hormones, we considered ACTH as a function of lagged cortisol levels over the time period of +40 to +180 min (Fig. 4). The graph suggests that polydipsic hyponatremics were not only relatively insensitive to cortisol but also shifted relative to the other three groups. Thus, at any given level of cortisol, ACTH levels were higher in polydipsic hyponatremics, although they seemed to decline at least at the highest levels of cortisol. Fitting a mixed-effects linear regression model again demonstrated that the slope of the relationship was not as steep in the hyponatremic polydipsics (Z = 2.28, P = 0.03) and was steeper in the nonpolydipsic patients, compared with the other three groups (Z = 2.37, P = 0.02).
|
Salivary cortisol
Salivary cortisol was measured from when subjects awoke to lights out on a separate occasion. All groups showed a similar steep drop in morning levels followed by slower declines in the afternoon and evening (Fig. 5, linear trend Z = 12.27, P < 0.001; quadratic trend Z = 5.36, P < 0.001).
|
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
). Furthermore, the relationship of ACTH to cortisol was shifted and blunted in the hyponatremic patients, whereas it was enhanced in the nonpolydipsic patient group (Fig. 4). The data did not support our second hypothesis (i.e. that the rate of ACTH decline in polydipsic groups will be blunted relative to healthy controls). This data analysis was complicated by the lower basal ACTH levels in the latter group. Together these findings support the previous observation that impaired glucocorticoid-negative feedback is found more commonly in polydipsic than nonpolydipsic schizophrenic patients, (4) but also support previous data that glucocorticoid regulation under physiologic conditions is not fundamentally altered (Fig. 5) (34).
The blunted response in the hyponatremic subjects cannot be attributed to recognized factors. Thus, preinfusion cortisol levels were similar across groups, and basal ACTH levels were similar in the three patient groups (Fig. 3, A and B, and Table 2). Whereas cortisol levels increased more slowly in the hyponatremic subjects, peak cortisol levels did not appear to differ, controlling for the rate of cortisol rise did not alter the findings, and the relationship between ACTH and cortisol was also blunted. A diurnal shift in HPAA axis sensitivity can be discounted because patients were on the same routine for 6 wk before the study and diurnal salivary cortisol levels did not differ (Fig. 5). Recent data suggest atypical antipsychotic agents blunt cortisol activity (35), but the number of subjects taking these drugs did not differ across patient groups nor did excluding those subjects appreciably alter the findings. Other possible factors such as age (29), obesity (26), acute psychiatric symptoms or levels of stress, nausea, or lightheadedness can be discounted because groups resembled each other on these measures (Tables 1 and 2). Whereas we did not include a group of nonpsychiatric subjects with hyponatremia or polydipsia, water imbalance per se seems an unlikely explanation because chronically hyponatremic animals have normal glucocorticoid feedback (36). The blunted ACTH sensitivity to cortisol in the hyponatremics and the enhanced sensitivity in nonpolydipsic patients are presumably offset by comparable differences in cortisol sensitivity to ACTH, because otherwise diurnal cortisol levels should have differed (12). Our unpublished data support this line of reasoning.
Our findings provide additional support for the view that neuroendocrine defects in hyponatremic schizophrenic patients (4, 7, 8) are a product of hippocampal-mediated mineralocorticoid dysfunction. Hence, this study was conducted during the diurnal nadir when stimulatory drive is minimal and hippocampal mineralocorticoid receptors modulate the HPAA axis (12, 25, 26, 27, 28). Indeed, the apparent shift in the relationship of ACTH and cortisol in the hyponatremics is consistent with mineralocorticoid dysfunction (12, 13) and is reminiscent of the unexplained shift in the relationship between vasopressin and plasma osmolality seen in these patients (7). The hippocampus (10, 11), in part via mineralocorticoid receptors (37), also normally constrains HPAA and vasopressin responses to psychological stimuli. Previous studies also demonstrate that hyponatremic patients exhibit both smaller anterior hippocampal formations (14) and greater HPAA and arginine vasopressin responses to psychological (but not systemic) stress than matched patients or healthy controls (5). Indeed, disruption of the development of this hippocampal segment produces an animal model of schizophrenia (15) characterized by enhanced HPAA and vasopressin responses to psychological stress (17) as well as impaired glucocorticoid-negative feedback (16). Finally, the hippocampus and mineralocorticoid receptors contribute to basic cognitive functions, (12) including the ability to behaviorally adapt to psychological stress (15, 38), which is also impaired in many schizophrenic patients (39). Thus, although speculative at this juncture, this subset of schizophrenic patients may provide unique opportunities to clarify the role of the hippocampus and mineralocorticoid receptors in both neuroendocrine regulation and severe mental illness.
There are several strengths of this study. Inclusion of matched psychiatric patient control groups diminishes the likelihood that findings in hyponatremics could be a consequence of nonspecific factors associated with schizophrenia, psychotropic medications, or polydipsia. Furthermore, psychiatric subjects were well characterized, habituated to the research setting, and maintained on a common diurnal cycle. The clinical research center setting provided for accurate sampling and measuring of potential confounding factors. Conducting the study in the evening after metyrapone treatment helped assure that changes in ACTH were a consequence of cortisol actions on mineralocorticoid receptors (25, 26, 29). Studying healthy controls twice enabled us to precisely characterize the normal ACTH response under these conditions, and measurement of diurnal cortisol activity enabled us to eliminate the possibility that changes in diurnal sensitivity were responsible for the findings.
Several limitations, however, also require emphasis. As in our previous studies (6, 7, 8), sample size was small, raising the issue of the robustness of the findings and whether undetected differences could have accounted for the results. Furthermore, by restricting ourselves to three a priori contrasts, we could not explore all group differences. Subjects with water imbalance were neither as polydipsic nor as hyponatremic as in previous studies (7, 8). Whereas this should have reduced differences between groups, it might also increase the possibility that undetected differences account for the findings. By not including a diluent infusion in the psychiatric patients, we cannot be absolutely certain that differences in ACTH responses are a consequence of differences in the effects of cortisol. Most of the data cited above regarding the functional role of hippocampus come from studies of rodents, not humans or other primates. Finally, whereas the enhanced feedback in the nonpolydipsic patients essentially eliminates the possibility that findings in hyponatremics are a nonspecific result of schizophrenia or its treatment, its etiology and significance remain to be determined.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Disclosure Statement: The authors have nothing to disclose.
First Published Online November 21, 2006
Abbreviations: HPAA, Hypothalamic-pituitary-adrenal axis; PANSS, Positive and Negative Symptom Scale.
Received May 25, 2006.
Accepted November 9, 2006.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
) or diluent (•) as illustrated in Fig. 1
), polydipsic and normonatremic (PNS, 
), or nonpolydipsic and normonatremic (NNS, •), and healthy controls (
