Further Evidence for a Specific Role of Estradiol in Schizophrenia? Anita Riecher-ROssler, Heinz H ner, Angelika Diitsch-Strobel, Manfred Oster, Michael Sturnbaum, Martina van Gtilick-Bailer, and Walter L6ffler Key W o r d s : Estradiol, menstrual cycle, psychopathology, schizophrenia, depression
Introduction There is a growing body of evidence for a specific role of estrogens in schizophrenia (for reviews see Seeman and Lang 1990; RiecherR~ssler and Hiifner 1993). In own systematic investigations we have shown that estradiol seems to have a protective effect in schizophrenic disorders: in a large epidemiological study we found that women in comparison with men, not only have a delayed peak of onset, but also exhibit an additional, smaller peak after age 40-45 (HMncr et al 1991, 1992). H~ifner et al postulated that this could be due to a rise in the vulnerability threshold for the outbreak of the disease, caused by estrogen, an effect that would decrease as estradiol serum levels fell in premenopause/menopause, which occurs around the same age. In animal experiments he found evidence that chronic estradiol treatment reduces the sensitivity of dopamine D2-receptors in the brain (Hiifner et al 1991). In a clinical study we were then able to show that estrogens seem to ameliorate psychopathology in schizophrenic women. Their symptomatotogy could be shown to vary with estradiol serum levels throughout the menstrual cycle. It was more likely to improve when estradiol levels rose and vice versa. The only exception was found in depressive symptoms. And as we have shown, the association was not caused by neuroleptics as an intermediate factor (Riecher-ROssler et al 1994). In line with this finding, Hallonquist et al (1993) observed lower psychopathology scores in schizophrenic women during the midluteal phase, as compared to the early follicular phase of the menFrom the Central Instituteof Mental Health, Mannheim,Germany. Address reprint requests to Dr. Anita Riecher-R6ssler, Central Institute of Mental Health, P.O. Box 122 120,68072 Mannheim,Germany. This study was supported by the DFG (German Society for the Advancement of Scientific Research) and the BMFT (Federal Ministry of Research and Technology.) Received May 11, 1993; revised March 18, 1994.
© 1994 Society of Biological Psychiatry
strual cycle. Although this association is an indirect one--estradiol blood levels were not measured---and only based on five nonacute outpatients, they concluded that estrogens may act as "endogenous neuroleptics." The question is, are these results specific for schizophrenia, however? In a further clinical study presented here, we have now tested this question of specificity by examining an age-matched control group of 29 women with acute psychiatric disorders other than schizophrenia. The results of the schizophrenic women will be given for comparison.
Methods We examined 32 Caucasian women acutely admitted (consecutive admissions to Pfalzidinik Landeck from June 1988 to April 1989) with a diagnosis of schizophrenia (PSE/CATEGO, Wing et all 1973) aged 30.5 (SD 6,5; range ! 8--43) and 29 age-matched (mean 31.2; SD 5,8, range 22--43) Caucasian women who acted as controis, acutely admitted to the Central Institute of Mental Health between October ! 989 and October 1992 for a psychiatric disorder other than schizophrenia: 24 cases of major depressive disorder: 1 case of bipolar disorder; 3 cases of adjustment disorder (in 2 cases with suicidal attempts): 1 case of panic disorder (DSM-IIIR/SCID, Spitzer et al 1989). All of these control patients had a marked depressive symptomatology. Excluded were patients with endocrine disease, patients who used hormonal contraception, or who had irregular cycles. Mean duration of disease since first psychiatric admission was 2.5 years in both groups (0-14 years in schizophrenic and 0-22 years in control women). Each woman was examined on admission and during her inpatient stay on days 2, 7, 13, 14, 21, and 28 of her menstrual cycle (mean number of assessments: 5.5 (SD i,5; range 4-9) in schizophrenics and 6.5 (SD 2,8; range 2-15) in controls, using the Brief Psychiatric Rating Scale (BPRS, Overall and Gorham 1962), the 0006-3223/94/$07.00
Brief Reports
~PS~CmAT~V
493
Table I. MeanPsychelmhologyScocesonAdmissiom
Schizoplu~mics (n = 32)
Parameters
BPRS-tolal score BPRS subscores anxiety/depression anergia thought disturbance activation hostile-suspiciousness NOS[E-total scor~ BfS-total score PDS-paranoid subscore
PDS-depression subscore
mean
(SD)
70,5
(15,0)
43,0
~,7~
13,0 10,8 17.4 14,8 12,7 183,5 26,9 20.9 I 7,7
(4,9) (4~2) (6,6) (4.4) (5,0) (31,I) (16,6) (| I+8) ( 12.i )
I "/,2 8,8 5,0 T,5 4,6
(4,0) (3~) (1,9) (2,8) ( i +9)
210,8
(19,6)
38,4 5,I 23,2 23,6
(163) (3,8) (10.7) (8,4)
--not evaluated--
HAMD-total score
m¢~
(~)
"Conlrar~to the ethersc~es, in ~ totalNOSIEsco~ a lugbervaluemeansle~ psychc~athoI~gy.
Nurses" Observation Scale for Inpatient Evaluation (NOSIE, Honigfeld et al 1976), the Paranoid-Depressivilfits-Skalaand the Befindlichkeits-Skala (PDS and BfS, v. Zerssen and Koeller ! 976) --self-rating scales for paranoid/depressive tendencies and general well-heing--and the Hamilton Depression Scale (HAMD, Hamilton 1976) for the controls. Raters assessing psychopathology were blind to the patients" cycles, patients were not aware of our hypothesis. On each assessment day, blood was taken in the morning before medication (around 8 AM), serum was kept frozen at - 4 0 °. To avoid methodological variations, all analyses (immuno-assay J- 125 esiradiol dir. Baxter) were carried out in one single batch. In order to determine the association of estradiol blood levels with psychopathology scores throughout the menstrual cycles, cross-correlations (Jenkins and Watts ! 968) were calculated. This method is well established in scientific electroencephalogram (EEG) analysis for measuring the similarity between two curves. According to this method correlations are calculated not interindividually but intraindividually. In our case it means that we CalCU14gitqk,,qkA~ , I U , ~ " q ~ J l l ~ l C ~ U I U C l l , ~
i q L t l v~Cg&~ll | l l 0 [ , ~ l W l U U C g & l ~gltl~dg ~ [ i i ~ l l
~klLFI~
[~w~[~ltgZ-
diol × each psychopathology score), thus getting 10 association parameters for each schizophrenic and I 1 association parameters for each control patient. Then for each sample and each association parameter we calculated Wilcoxon tests over the group as a whole to determine the significance of the associations.
Results For information on the severity of disease, Table 1 shows the various psychopathology scores on admission. The results concerning the association between estradiol serum levels and psychopathology scores over time are reported in Table 2. For each psychopathology score the table shows the means of the 32 schizophrenics" and the 29 controls" individual correlation coefficients.
Discussion It is striking to see that in schizophrenic women almost all psychopathology scores correlated significantly with the estradiol blood
level, but this was not so in the control group. In schizop~a~nic women, symptoms were more likely to improve when estradiol levels rose and vice versa. We have interpreted this finding as further evidence for a protective effect of estradiol in schizophrenia, possibly due t~ the presumable antidopaminergic activity of this hormone, which has been shown in many recent s t ~ (for reviews see Seeman and Lang |990; Riechex-Rtssler and H~uer 1993). The fact that estradiol did not seem to influence depressive sym~omatology in schizophrenics indicated a relative specificity of this eff~t. In the present study we have further tested the question of specificity. The results show that in an age-matched control group, which mainly consisted of women with depressive syndromes, symptomatology was not correlated with estradiol blood levels. The only exception was "general behavior on ward." If only the control patients with major depressive disorders were taken into account, the results remained essentially unchanged: only "behavior on ward" was associated with estradiol blood levels (p < 0.05). Thus the clinical effect of estrogens in fact resembles that of ncuroleptics: Tney seem to ameliorate psychotic symptomatology ~ possibly also general behavior, but not depressive symptomatology. Although we found a significant association between estradiol levels and most psychopathology scores in the schizophrenic, but not in the control patients, the differences between the two groups, however, were statistically not significant. The reason for this is a relatively large overlap between the two groups, because some individuals in the control group showed similar associations concerning some psychopathology scores as did the schizophrenic women. Fm~,her analyses will he conducted to identify these patients and their characteristics. There are also some limitations to our study, for example, because of ethical reasons we could not leave the acutely ill schizophrenic women without neuroleptics for such a long period. Also in the control group three patients had to be treated with neuroleptics during their hospital stay. But as we have shown elsewhere (Riecher-ROssler et al 1994), the association found in schizophrenic women was not caused by neuroleptics as an intermediate factor.
.IOL PSYCHIATRY
4~
Bl'icf
I~M;~:492-.494
Table 2, Correlations between Psychopathology Scores and Estradiol Serum Levels throughout Hospital Stay of Schizophrenic and Nonschizophrenic Women. Nonschi~
Schizophrenics (n = 32) Parameters
n~
m~m
(SD)
p
nb
mean
BPRS-totalscore BPRS subscores anxiety/depression anergia thought disturbance activation hostile-suspiciousness NOSIE-totalscore¢ BfS-total score PDS-paranoidsubscore PDS-depressionsubscore HAMD-totalscore
31
-0'25
(0.4 I)
< 0;01
29
(0,52) (0,42) (0.44) (0,4 ! ) (0.47) (0.49) (0.43) (0,42) (0,52)
NS < 0,10 < 0,01 < 0,01 < 0,10 < 0,01 < 0,05 <.0.05 NS
29 28 15 29 22 29 29 27 29 29
31 31 31 31 31 32 32 32 32
-0,10 -0,15 -0,28 -0,27 -0,19 0,25 -0,20 -0,17 -0,10 - - not evaluated--
(n = 29) (SD)
p
0,03
(0.55)
NS
O,O0 -0,04 -0A3 -0,,86 -0,13 0,22 -0,02 ~ 0,01 0,03
(0,52) (0,53) (0.55) (0,49) (0.45) (0.43) (0,56) (0,5 I) (0.56) (0,53)
NS HS
NS NS NS < 0.01 NS NS NS NS
-Shownas meansof individualcorrelationcoefficients. bTheslightlyvaryingnumbersare partlydue to missingdata,partlydue to exclusions:patients,whonevershowedsymptomsof a certainscoreo~nevershowedv~riabifilyin a ~rtam scorewereexcludedfromarab.sisconcerningthe resp.score( 14 controlsconcerning"thoughtdistucbance."6 controls¢onoeming" h o s b l e - ~ " and 2 controlsconcerning"PDS-paranoidsubscore"), '-Contraryto theotherscores,in the totalNOSIEscorea highervaluemeanslesspsychopathology. If our results can be replicated they could have some interesting implications: As regards research, they seem to give further evidence of a psychotropic potential of estrogens. As regards the clinical practice, they could offer new therapeutic strategies: in young fertile women, one could try to adjust the neuroleptic dosage to the menstrual cycle---at least in those showing perimenstrual deterioration of symptoms. In schizophrenic women before and after menopause, one could substitute estrogens, a measure
that has already proved to have a beneficial effect in women of this age-group for organic reasons, for example, cardiovascular disease or ostooporosis (Henderson et al 1991; Kuhl 1993). Corresponding intervention studies should be undertaken. We are grateful to Dr. Christi Jennen-Steinnw.tzand Andw.asWolffor their statistical help.
References H~ifner H, Behrens S, de Vry J, Gattaz WF (1991 ): Schizophrenia: Gender and age. Why is the onset of schizophrenia later in women? Biol Psychiatry 1:431-433. Hafner H, Riecher-ROssler A, Maurer K, F~itkenheuer B, LOftier W (1992): First onset and early symptomatology of schizophrenia. A chapter of epidemiological and neurobiological research into a~e and ~ x difft,a'~nrP_~ Fur Arch P~ychiat,~ Clin Neurosci 242:109-118. Hallonquist JD, Seeman MV, Lung M, Rector NA (1993): Variation in symptom severity over the menstrual cycle of schizophrenics. Blot Psychiatry 33:207-209. Hamilton M (1976): Hamilton Depression Scale. In Guy W (ed),
ECDEU Assessment Manual for Psychopharmacology, rev. ed. Rockville, MD, NIMH, pp 179-192. Henderson B ~ Paginini-Hill A, Ros RK (1991): Decreased mortality in users of estrogen replacement therapy. Arch Intern Med ! 51:75-8. Honigfeld G, Gillis RD, Klett CJ (1976): NOSIE: Nurses' observation scale for inpatient evaluation. In Guy W (ed), ECDEU Assessment Manual for Psychopharmacology, rev. ed. Rockville, MD: NIMH. Jenkins GM, Watts DG ( 1968 ): SpectralAnalysis and its Applications. Oakland, CA: Holden Day. Kuhl H (1993): Substitution mit Ostrogenen und Gestagenen.
Kongressbericht vom 9. Arbeitstreffen des Ziircher Gesp-
r~chskreises. Deutsches Ar-zleblatt90:1532-1533. Overall JE, Gorham DR (1962): The Brief Psychiatric Rating Scale. Psychol Rep 10:799-812. Riecher-R0ssler A, I-I'~ifnerH (1993): Schizophrenia and oestrogens--is there an association? Eur Arch Psychiatry Clin Neurosci 242:323-328. Riecher-Rtssler A, H&'ner H, Stumbanm M, Manrer K, Schmidt R (1994): Can estradiol modulate schizophrenic symptomatology?SchizophrBull, ! :203-214. Seeman MV, Lung M (1990): The role of estrogens in schizophrenia gender differences. Schizophr Bull 1 6 : 1 8 5 - 1 9 4 . Spitzer RL, Williams JB, Gibbon M (1989): Structured Clinical Interview for DSM-II! R: SCID. New York: New York State Psychiatric Institute, Biometrics Research Division (German version by Wittchen et a11989). v. Zers~n D, Koeller DM (1976): Klinische Selbstbeurteilungs-
Skalen (KSb-Si) aus dem Miinchener Psychiatrischen lnformations-System (PSYCHIS Miinchen) Manuale. Weinheim: Beltz. Wing JK, Cooper JE, Sartorius N (1973): Measurement and Classification of Psychiatric Symptoms. London: Cambridge University Press.