European Neuropsychopharmacology, 2 (1992) 99-106 !~ 1992 Elsevier Science Publishers B.V. All rights reserved 0924-977X/92/$05.00
99
ENP 00053
HPA-related CSF neuropeptides in suicide attempters Lil T r ~ s k m a n - B e n d z a, R o l f E k m a n b, G 6 r a n Regn611 a a n d R o l l {Shman a ~Department (~l"Psychiatry, University Hospital, S-221 85 Lund, Sweden, and bDepartment q[ Psychiatry and Neurochemistry, St. Lars Hospital, Box 638, S-220 06 Lund, Sweden (Received 4 July, 1991) (Revised, received 7 January, 1992) (Accepted 22 January, 1992)
Key words'." Suicidal behavior; Major depression; Neuropeptides; Corticotropin-releasing hormone; Somatostatin; Delta-sleep-inducing peptide; Neuropeptide Y; Beta-endorphin; Vasopressin
Summary Corticotropin-releasing hormone (CRH), somatostatin (SOM), delta-sleep-inducing peptide (DSIP), neuropeptide Y (NPY), beta-endorphin (beta-END), and vasopressin (AVP), which are regarded as being involved in the HPA-regulation were investigated in lumbar CSF of 44 suicide attempters. The patients were diagnosed according to the DSM-III-R, and rated with the MADRS. The neuropeptides were compared with the serotonin metabolite 5-hydroxyindoleacetic acid (5-HIAA) in CSF and with post-dexamethasone plasma cortisol. We found strong correlations between CRH and the peptides SOM and beta-END. The latter also correlated positively with SOM. There were no differences between men and women. Patients with major depressive disorders had significantly lower SOM, CRH, and DSIP than other patients. Both SOM and betaEND correlated negatively with post dexamethasone plasma cortisol in all patients. We found no significant relationships between neuropeptides and CSF 5-HIAA. Patients who had made previous suicide attempts had significantly lower CRH than those who had not. No other significant associations between neuropeptides and suicidal subgroups of patients appeared, and there was no indication of specific neuropeptide patterns in patients who later completed suicide. Intercorrelations of some neuropeptides and low SOM and DSIP in major depressed patients are findings in line with those by others.
Introduction Several lines of evidence suggest that suicidal patients with major depression suffer from an abnormal function in the hypothalamic pituitary adrenocortical (HPA) axis (Bunney and Fawcett, 1965; Carroll et al., 1981; Krieger et al., 1971; Norman et al., 1990; Ostroff et al., 1982). It is generally believed that the abnormality of the HPA Correspondence to: Dr. Lil Tr~skman-Bendz, Department of Psychiatry, University Hospital, S-221 85 Luud, Sweden. Tel.: 46173840, Fax no: 46110923.
axis is localized in the hypothalamic-pituitary region, though a hyperresponsiveness to corticotropin (ACTH) of the adrenal cortex has also been discussed (Gold and Rubinow, 1987). Initially, the dexamethasone test (DST) was thought to be a specific HPA test for endogenous depression (melancholia) (Carroll, 1982). Later, the test was proved to be nonspecific, since a blunted cortisol response was also seen in patients with other psychiatric disorders (Berger et al., 1984). Based on the assumption that lumbar cerebrospinal fluid (CSF) measures of neuropeptides reflect activity in central peptidergic neurons (Burbach,
100 1982), several CSF studies have been performed in healthy subjects, and in patients suffering from psychiatric illnesses. Immunoreactive corticotropin-releasing hormone (CRH) and somatostatin (SOM) correlate in healthy subjects and in euthymic bipolar patients. This could be indicative of SOM acting as a neuromodulator on hypothalamic CRH, or the reverse (Berrettini et al., 1987a; Berrettini et al., 1988). Somatostatin is known to be inhibitory, while C R H is excitatory. Furthermore, these neuropeptides have a modulatory effect on monoamines (Gold and Rubinow, 1987). The coexistence in neurons of peptides and biogenic amines is now generally acknowledged (H6kfelt et al., 1977). Low concentrations of SOM have been reported in depressed patients, patients with dementia of the Alzheimer type, and in patients with anorexia nervosa. However, these findings are not consistent. SOM levels have also been reported to be low in a variety of psychiatric patients, who did not suppress plasma-cortisol after dexamethasone. SOM levels are not reported to be low in healthy volunteers, or in euthymic affectively ill patients. (Agren and Lundqvist, 1985; Doran et al., 1986; Minthon et al., 1990; Rubinow, 1986). C S F - C R H did not differ from controls in schizophrenic and demented patients, but depressed patients had high concentrations (Banki et al., 1987; Kaye et al., 1987; Nemeroffet al., 1984). Roy et al. (1987) found no C R H differences between depressed patients and healthy controls. They suggested an association between high C R H and high post dexamethasone plasma cortisol in depressed patients. In patients with dementia, there was an association between low C R H in CSF and cognitive impairment (Pomara et al., 1989). A significantly positive correlation between C R H and the serotonin metabolite 5-hydroxyindoleacetic acid (5-H1AA) was found in spinal fluid of depressed and demented patients (Pomara et al., 1989; Widerl6w, 1985). Similarly, another study showed an association between CSF cortisol and CSF 5-HIAA in depressed patients, but not in healthy controls (Trfiskman et al., 1980). Roy et al. (1986) suggested that depressed patients who had low CSF concentrations of 5HIAA, and who were non-suppressors of cortisol after dexamethasone, were at risk for future suicidality. These findings are important, since the serotonergic system is involved in suicidal and impulsive behavior (Asberg et al., 1987). The delta-sleep-inducing peptide (DS1P) has
been described to influence plasma levels of pituitary hormones (Bjartell et al., 1988). Furthermore, DSIP seems to be involved in monoaminergic transmission by modulating noradrenergic receptors. Decreased levels of DSIP have been reported in plasma and CSF of depressed patients, and patients with schizophrenia (Lesch et al., 1988: Wall~us et al., 1984). DSIP might also influence alertness and coping-ability (Schneider-Helmert and Schoenenberger, 1983). Neuropeptide Y (NPY) is among the most abundant neuropeptides in the CNS (Allen et al., 1983). NPY administration to experimental animals produces increases of C R H in the hypothalamus (Haas and George, 1987). SOM neurons coexist with NPY neurons in hippocampus and cerebral cortex (Chan-Palay, 1987). Since its recent discovery, NPY has been attributed to a variety of functional roles, including the pathogenesis of depressive disorder. However, CSF findings of NPY in psychiatric patients are not consistent (Atack et al., 1988; Berrettini et al., 1987b; Widerl6w et al., 1988a). Lindstr6m (1985) reported a significantly inverse correlation between CSF NPY and 5-HIAA in schizophrenic patients and healthy controls. Beta-endorphin (beta-END) and ACTH can arise from the same prohormone proopiomelanocortin (POMC) in the pituitary. Therefore, betaE N D like immunoreactivity has been studied in plasma before and after dexamethasone administration in order to investigate its role in affective disorders. There are also CSF studies of betaEND. The findings so far, are however far from conclusive (Berger and Nemeroff, 1987). The aim of the present investigation, which is part of a comprehensive study of suicide attempters, was to examine the pattern of some CSFneuropeptides related to the HPA axis in suicide attempters, and their relationship with CSF 5HIAA.
Experimental procedures Subjects Patients who had made a suicide attempt were recruited from the emergency room, or from the intensive care unit immediately after the attempt. They were all in-patients on a specialized ward. They gave written informed consent to participate in our research program.
101
Psychopathology Our patients were diagnosed according to the DSM-III-R (1987) by two independent psychiatrists. Patients could have more than one axis Idiagnosis. On the day of the spinal tap, they were rated according to the Comprehens!ve Psychopathological Rating Scale (CPRS) (Asberg et al., 1978). In our calculations, we used the Montgomery-Asberg Depression Rating Scale (MADRS), which was extracted from the CPRS (Montgomery and Asberg, 1979). Suicide attempts We regard an attempted suicide as a lifethreatening behavior with the intent of jeopardizing one's own life. Suicide attempts were classified as violent or non-violent (Tr~iskman et al., 1981). Violent attempters had used methods other than drug overdosages or single wrist cuts. They could also have used combinations of methods. We also subdivided our patients into those who had and those who had not used single or several cuts at index or previous suicide attempts. Patients who had made previous suicide attempt(s) were denoted repeaters. Suicide-related ratings The suicide intent of our patients was rated according to the Suicidal Intent Scale (SIS), which was constructed by Beck et al. (1974a). Shortly after admission to the ward, our patients filled out a self-rating scale, the Hopelessness scale (HS) (Beck et al., 1974b), which has been shown to be predictive of future suicide by Beck et al. (1990).
Methods
Lumbar puncture Lumbar punctures were performed after a washout period of 16 _+ 7 days from admission to the psychiatric research ward. Some patients had stayed for several days in the intensive care unit, or in the psychiatric emergency unit before admission to the research ward. Patients who had taken overdoses of antidepressant or neuroleptic compounds had longer wash-out periods (19.0 _+ 7.08 days) than those who had taken other compounds or used other methods (15.5 _+ 7.19 days; t = 1.42, N.S.). Occasional doses of benzodiazepines were allowed during the wash-out period. In the morning of the spinal tap, blood was
taken for plasma blanks, to check for the presence of psychotropic drugs. All samples were blank. The tap was performed in the morning between 8 and 9 a.m. after one night of fasting and bedrest. Patients were sitting. Twelve + 6 ml CSF was taken between L4 and L5. The CSF was immediately centrifuged and stored in 2-ml portions at - 8 0 ° C until assayed.
Biochemical assays Neuropeptides Neuropeptides were determined in the 6 ml CSF portion and by radioimmunoassays (RIA). In the hCRH-RIA, we used a rabbit antiserum raised against h C R H conjugated to bovine serum albumin with glutaraldehyde. The antiserum was used in a final dilution 1:80000. mSI-hCRH was prepared by the chloramine-T method and purified by high-performance liquid chromatography (HPLC). The antiserum does not cross-react with o C R H ( I ~ I ) , hCRH(34~41), hCRH-(37~I), adrenocorticotrophic hormone (ACTH), or beta-END. The detection limit is 2 pmol/1. The interassay variation expressed as CV was 12%. For the RIA of beta-END a rabbit antiserum (K-7762), used in a previously described betaE N D - R I A (Bach et al., 1986, 1987), recognizes the N-terminal portion and does not cross-react with beta-lipotropin. The detection limit was 20 pmol/1. The interassay variation expressed as CV was 15%. Immunoreactive DSIP was measured by a specific RIA procedure with a detection limit of 90 pmol/l. The interassay coefficient was below 15% (Ekman et al., 1983). Immunoreactive neuropeptide Y (NPY) was measured using a rabbit antiserum raised against synthetic porcine N P Y (Emson, Cambridge, U.K.). The antiserum cross-reacted with human NPY to 100%, NPY(2 36) 33%, NPY(4-36) 29%, peptide YY (PYY) to 33%, but not with C-terminal fragments of NPY and PYY. The detection limit was 25 pmol/1 and the interassay variation expressed as CV < 12% (Widerl6w et al., 1988a). Immunoreactive somatostatin (SOM) was determined using a rabbit antiserum (K18 Milab, Malm6, Sweden), in a final dilution of 1:25 000. It does not cross-react with any other known neuropeptides besides cyclic SOM (100%), linear SOM (50%), (tyrl)-SOM (100%). The detection limit is 5 pmol/l, expressed as SOM15 28 ekv, the interassay coefficient of variation being < 1 2 % (Wallgren et al., 1987). Immunoreactive arginine vasopressin (AVP)
102 was measured by a previously described RIA (Widerl6w et al., 1988b). The antiserum showed little cross-reaction with related peptides (lysine vasopressin < 0 . 0 3 % ; arginine vasotocine; oxytocine < 0 . 0 1 % ) . The limit o f detection of plasma AVP after ethanol extraction was 0.4 pmol/1 and the interassay variation < 9%. 5 - H I A A / C S F A mass-fragmentographic m e t h o d was used (Swahn et al., 1976) and 5-HIAA was measured in the 12-ml CSF portion. Dexamethasone test Samples for plasma cortisol were taken at 4 p.m. on the day o f the spinal tap, before 1 mg dexamethasone was given at 11 p.m., as well as at 9 a.m. and 4 p.m. the following day. In our calculations we only used the 4 p.m. postdexamethasone cortisol. Statistical methods When performing group comparisons we used Student's t-test, M a n n - W h i t n e y and a two-way analysis o f variance (ANOVA). In the latter, actual values o f neuropeptides were replaced b y ranks (Conover and Iman, 1981). F o r correlations we used Spearman's rho. Ethics Committee This study was approved by the Lund University Medical Ethics Committee.
Results
Patients Data were available from 44 patients, whose age was 38.75 _+ 13.05 years. There were 24 women (41.4 ± 14.4 years) and 20 men (35.6 ± 10.7 years) (t = 1.51, N.S.). Diagnostic categories Sixteen patients (10 women, 6 men) were diagnosed as having major depressive disorder ( M D D ) , 9 with melancholia. M D D patients were significantly older than n o n - M D D patients (43.75 _+ 12.92 years vs 34.82 _+ 11.48 years, t 2.37, P
99
ENP 00053
HPA-related CSF neuropeptides in suicide attempters Lil T r ~ s k m a n - B e n d z a, R o l f E k m a n b, G 6 r a n Regn611 a a n d R o l l {Shman a ~Department (~l"Psychiatry, University Hospital, S-221 85 Lund, Sweden, and bDepartment q[ Psychiatry and Neurochemistry, St. Lars Hospital, Box 638, S-220 06 Lund, Sweden (Received 4 July, 1991) (Revised, received 7 January, 1992) (Accepted 22 January, 1992)
Key words'." Suicidal behavior; Major depression; Neuropeptides; Corticotropin-releasing hormone; Somatostatin; Delta-sleep-inducing peptide; Neuropeptide Y; Beta-endorphin; Vasopressin
Summary Corticotropin-releasing hormone (CRH), somatostatin (SOM), delta-sleep-inducing peptide (DSIP), neuropeptide Y (NPY), beta-endorphin (beta-END), and vasopressin (AVP), which are regarded as being involved in the HPA-regulation were investigated in lumbar CSF of 44 suicide attempters. The patients were diagnosed according to the DSM-III-R, and rated with the MADRS. The neuropeptides were compared with the serotonin metabolite 5-hydroxyindoleacetic acid (5-HIAA) in CSF and with post-dexamethasone plasma cortisol. We found strong correlations between CRH and the peptides SOM and beta-END. The latter also correlated positively with SOM. There were no differences between men and women. Patients with major depressive disorders had significantly lower SOM, CRH, and DSIP than other patients. Both SOM and betaEND correlated negatively with post dexamethasone plasma cortisol in all patients. We found no significant relationships between neuropeptides and CSF 5-HIAA. Patients who had made previous suicide attempts had significantly lower CRH than those who had not. No other significant associations between neuropeptides and suicidal subgroups of patients appeared, and there was no indication of specific neuropeptide patterns in patients who later completed suicide. Intercorrelations of some neuropeptides and low SOM and DSIP in major depressed patients are findings in line with those by others.
Introduction Several lines of evidence suggest that suicidal patients with major depression suffer from an abnormal function in the hypothalamic pituitary adrenocortical (HPA) axis (Bunney and Fawcett, 1965; Carroll et al., 1981; Krieger et al., 1971; Norman et al., 1990; Ostroff et al., 1982). It is generally believed that the abnormality of the HPA Correspondence to: Dr. Lil Tr~skman-Bendz, Department of Psychiatry, University Hospital, S-221 85 Luud, Sweden. Tel.: 46173840, Fax no: 46110923.
axis is localized in the hypothalamic-pituitary region, though a hyperresponsiveness to corticotropin (ACTH) of the adrenal cortex has also been discussed (Gold and Rubinow, 1987). Initially, the dexamethasone test (DST) was thought to be a specific HPA test for endogenous depression (melancholia) (Carroll, 1982). Later, the test was proved to be nonspecific, since a blunted cortisol response was also seen in patients with other psychiatric disorders (Berger et al., 1984). Based on the assumption that lumbar cerebrospinal fluid (CSF) measures of neuropeptides reflect activity in central peptidergic neurons (Burbach,
100 1982), several CSF studies have been performed in healthy subjects, and in patients suffering from psychiatric illnesses. Immunoreactive corticotropin-releasing hormone (CRH) and somatostatin (SOM) correlate in healthy subjects and in euthymic bipolar patients. This could be indicative of SOM acting as a neuromodulator on hypothalamic CRH, or the reverse (Berrettini et al., 1987a; Berrettini et al., 1988). Somatostatin is known to be inhibitory, while C R H is excitatory. Furthermore, these neuropeptides have a modulatory effect on monoamines (Gold and Rubinow, 1987). The coexistence in neurons of peptides and biogenic amines is now generally acknowledged (H6kfelt et al., 1977). Low concentrations of SOM have been reported in depressed patients, patients with dementia of the Alzheimer type, and in patients with anorexia nervosa. However, these findings are not consistent. SOM levels have also been reported to be low in a variety of psychiatric patients, who did not suppress plasma-cortisol after dexamethasone. SOM levels are not reported to be low in healthy volunteers, or in euthymic affectively ill patients. (Agren and Lundqvist, 1985; Doran et al., 1986; Minthon et al., 1990; Rubinow, 1986). C S F - C R H did not differ from controls in schizophrenic and demented patients, but depressed patients had high concentrations (Banki et al., 1987; Kaye et al., 1987; Nemeroffet al., 1984). Roy et al. (1987) found no C R H differences between depressed patients and healthy controls. They suggested an association between high C R H and high post dexamethasone plasma cortisol in depressed patients. In patients with dementia, there was an association between low C R H in CSF and cognitive impairment (Pomara et al., 1989). A significantly positive correlation between C R H and the serotonin metabolite 5-hydroxyindoleacetic acid (5-H1AA) was found in spinal fluid of depressed and demented patients (Pomara et al., 1989; Widerl6w, 1985). Similarly, another study showed an association between CSF cortisol and CSF 5-HIAA in depressed patients, but not in healthy controls (Trfiskman et al., 1980). Roy et al. (1986) suggested that depressed patients who had low CSF concentrations of 5HIAA, and who were non-suppressors of cortisol after dexamethasone, were at risk for future suicidality. These findings are important, since the serotonergic system is involved in suicidal and impulsive behavior (Asberg et al., 1987). The delta-sleep-inducing peptide (DS1P) has
been described to influence plasma levels of pituitary hormones (Bjartell et al., 1988). Furthermore, DSIP seems to be involved in monoaminergic transmission by modulating noradrenergic receptors. Decreased levels of DSIP have been reported in plasma and CSF of depressed patients, and patients with schizophrenia (Lesch et al., 1988: Wall~us et al., 1984). DSIP might also influence alertness and coping-ability (Schneider-Helmert and Schoenenberger, 1983). Neuropeptide Y (NPY) is among the most abundant neuropeptides in the CNS (Allen et al., 1983). NPY administration to experimental animals produces increases of C R H in the hypothalamus (Haas and George, 1987). SOM neurons coexist with NPY neurons in hippocampus and cerebral cortex (Chan-Palay, 1987). Since its recent discovery, NPY has been attributed to a variety of functional roles, including the pathogenesis of depressive disorder. However, CSF findings of NPY in psychiatric patients are not consistent (Atack et al., 1988; Berrettini et al., 1987b; Widerl6w et al., 1988a). Lindstr6m (1985) reported a significantly inverse correlation between CSF NPY and 5-HIAA in schizophrenic patients and healthy controls. Beta-endorphin (beta-END) and ACTH can arise from the same prohormone proopiomelanocortin (POMC) in the pituitary. Therefore, betaE N D like immunoreactivity has been studied in plasma before and after dexamethasone administration in order to investigate its role in affective disorders. There are also CSF studies of betaEND. The findings so far, are however far from conclusive (Berger and Nemeroff, 1987). The aim of the present investigation, which is part of a comprehensive study of suicide attempters, was to examine the pattern of some CSFneuropeptides related to the HPA axis in suicide attempters, and their relationship with CSF 5HIAA.
Experimental procedures Subjects Patients who had made a suicide attempt were recruited from the emergency room, or from the intensive care unit immediately after the attempt. They were all in-patients on a specialized ward. They gave written informed consent to participate in our research program.
101
Psychopathology Our patients were diagnosed according to the DSM-III-R (1987) by two independent psychiatrists. Patients could have more than one axis Idiagnosis. On the day of the spinal tap, they were rated according to the Comprehens!ve Psychopathological Rating Scale (CPRS) (Asberg et al., 1978). In our calculations, we used the Montgomery-Asberg Depression Rating Scale (MADRS), which was extracted from the CPRS (Montgomery and Asberg, 1979). Suicide attempts We regard an attempted suicide as a lifethreatening behavior with the intent of jeopardizing one's own life. Suicide attempts were classified as violent or non-violent (Tr~iskman et al., 1981). Violent attempters had used methods other than drug overdosages or single wrist cuts. They could also have used combinations of methods. We also subdivided our patients into those who had and those who had not used single or several cuts at index or previous suicide attempts. Patients who had made previous suicide attempt(s) were denoted repeaters. Suicide-related ratings The suicide intent of our patients was rated according to the Suicidal Intent Scale (SIS), which was constructed by Beck et al. (1974a). Shortly after admission to the ward, our patients filled out a self-rating scale, the Hopelessness scale (HS) (Beck et al., 1974b), which has been shown to be predictive of future suicide by Beck et al. (1990).
Methods
Lumbar puncture Lumbar punctures were performed after a washout period of 16 _+ 7 days from admission to the psychiatric research ward. Some patients had stayed for several days in the intensive care unit, or in the psychiatric emergency unit before admission to the research ward. Patients who had taken overdoses of antidepressant or neuroleptic compounds had longer wash-out periods (19.0 _+ 7.08 days) than those who had taken other compounds or used other methods (15.5 _+ 7.19 days; t = 1.42, N.S.). Occasional doses of benzodiazepines were allowed during the wash-out period. In the morning of the spinal tap, blood was
taken for plasma blanks, to check for the presence of psychotropic drugs. All samples were blank. The tap was performed in the morning between 8 and 9 a.m. after one night of fasting and bedrest. Patients were sitting. Twelve + 6 ml CSF was taken between L4 and L5. The CSF was immediately centrifuged and stored in 2-ml portions at - 8 0 ° C until assayed.
Biochemical assays Neuropeptides Neuropeptides were determined in the 6 ml CSF portion and by radioimmunoassays (RIA). In the hCRH-RIA, we used a rabbit antiserum raised against h C R H conjugated to bovine serum albumin with glutaraldehyde. The antiserum was used in a final dilution 1:80000. mSI-hCRH was prepared by the chloramine-T method and purified by high-performance liquid chromatography (HPLC). The antiserum does not cross-react with o C R H ( I ~ I ) , hCRH(34~41), hCRH-(37~I), adrenocorticotrophic hormone (ACTH), or beta-END. The detection limit is 2 pmol/1. The interassay variation expressed as CV was 12%. For the RIA of beta-END a rabbit antiserum (K-7762), used in a previously described betaE N D - R I A (Bach et al., 1986, 1987), recognizes the N-terminal portion and does not cross-react with beta-lipotropin. The detection limit was 20 pmol/1. The interassay variation expressed as CV was 15%. Immunoreactive DSIP was measured by a specific RIA procedure with a detection limit of 90 pmol/l. The interassay coefficient was below 15% (Ekman et al., 1983). Immunoreactive neuropeptide Y (NPY) was measured using a rabbit antiserum raised against synthetic porcine N P Y (Emson, Cambridge, U.K.). The antiserum cross-reacted with human NPY to 100%, NPY(2 36) 33%, NPY(4-36) 29%, peptide YY (PYY) to 33%, but not with C-terminal fragments of NPY and PYY. The detection limit was 25 pmol/1 and the interassay variation expressed as CV < 12% (Widerl6w et al., 1988a). Immunoreactive somatostatin (SOM) was determined using a rabbit antiserum (K18 Milab, Malm6, Sweden), in a final dilution of 1:25 000. It does not cross-react with any other known neuropeptides besides cyclic SOM (100%), linear SOM (50%), (tyrl)-SOM (100%). The detection limit is 5 pmol/l, expressed as SOM15 28 ekv, the interassay coefficient of variation being < 1 2 % (Wallgren et al., 1987). Immunoreactive arginine vasopressin (AVP)
102 was measured by a previously described RIA (Widerl6w et al., 1988b). The antiserum showed little cross-reaction with related peptides (lysine vasopressin < 0 . 0 3 % ; arginine vasotocine; oxytocine < 0 . 0 1 % ) . The limit o f detection of plasma AVP after ethanol extraction was 0.4 pmol/1 and the interassay variation < 9%. 5 - H I A A / C S F A mass-fragmentographic m e t h o d was used (Swahn et al., 1976) and 5-HIAA was measured in the 12-ml CSF portion. Dexamethasone test Samples for plasma cortisol were taken at 4 p.m. on the day o f the spinal tap, before 1 mg dexamethasone was given at 11 p.m., as well as at 9 a.m. and 4 p.m. the following day. In our calculations we only used the 4 p.m. postdexamethasone cortisol. Statistical methods When performing group comparisons we used Student's t-test, M a n n - W h i t n e y and a two-way analysis o f variance (ANOVA). In the latter, actual values o f neuropeptides were replaced b y ranks (Conover and Iman, 1981). F o r correlations we used Spearman's rho. Ethics Committee This study was approved by the Lund University Medical Ethics Committee.
Results
Patients Data were available from 44 patients, whose age was 38.75 _+ 13.05 years. There were 24 women (41.4 ± 14.4 years) and 20 men (35.6 ± 10.7 years) (t = 1.51, N.S.). Diagnostic categories Sixteen patients (10 women, 6 men) were diagnosed as having major depressive disorder ( M D D ) , 9 with melancholia. M D D patients were significantly older than n o n - M D D patients (43.75 _+ 12.92 years vs 34.82 _+ 11.48 years, t 2.37, P
