Incidence of polycystic ovaries and androgen serum levels in women with borderline personality disorder

Journal of Psychiatric Research 44 (2010) 847–852

Contents lists available at ScienceDirect

Journal of Psychiatric Research journal homepage: www.elsevier.com/locate/jpsychires

Incidence of polycystic ovaries and androgen serum levels in women with borderline personality disorder Stefan Roepke a,*, Andreas Ziegenhorn a,c, Julia Kronsbein a, Angela Merkl a, Scharif Bahri a, Julia Lange b, Horst Lübbert b, Ulrich Schweiger d, Isabella Heuser a, Claas-H. Lammers a a

Charité-Universitätsmedizin Berlin, Campus Benjamin Franklin, Department of Psychiatry, Berlin, Germany Charité-Universitätsmedizin Berlin, Campus Benjamin Franklin, Department of Gynecology, Berlin, Germany Hôtel-Dieu Hospital, Department of Psychiatry, Saint-Jérôme, Québec, Canada d Department of Psychiatry, University of Luebeck, Lübeck, Germany b c

a r t i c l e

i n f o

Article history: Received 20 April 2009 Received in revised form 28 December 2009 Accepted 12 January 2010

Keywords: Borderline personality disorder Androgens Testosterone Androstenedione 17a-Hydroxyprogesterone Polycystic ovaries Depression PCO Glucose metabolism Obesity Visceral fat

a b s t r a c t Obesity, increased visceral fat and disturbed glucose metabolism have been found in borderline personality disorder (BPD) patients. These conditions are often associated with disturbed androgen metabolism. Elevated androgens in women are related to polycystic ovaries (PCO) and might have an impact on psychopathology. Thus, higher prevalence of PCO and elevated androgen levels are suspected in BPD. In the study, we examined 31 BPD patients and 30 healthy controls ultrasonographically for PCO and measured their serum levels of androgens and interacting hormones. Furthermore, influence on psychopathology of free testosterone (FT) serum level was assessed. PCO was significantly more prevalent in BPD patients (30.4%) compared to healthy controls (6.9%). Testosterone, FT, androstenedione (A), and 17a-hydroxyprogesterone (17-OHP) were significantly elevated in the BPD group independently of BMI. FT serum level significantly correlated with depressive symptoms. In summary, our data suggest a disturbed androgen metabolism in BPD patients. Ó 2010 Published by Elsevier Ltd.

1. Introduction Recent research on borderline personality disorder (BPD) has focused on comorbid medical conditions, especially obesity and obesity-related chronic medical disorders (Kahl et al., 2006; Frankenburg and Zanarini, 2006). Obesity and associated metabolic alterations appear to be linked to BPD as indicated by increased visceral fat and reduced insulin sensitivity in BPD patients (Kahl et al., 2005). Female obesity is often associated with profound alterations in androgen metabolism (Pasquali, 2006), which might have corresponding psychopathological effects (Archer, 2006). Women with central obesity have lower sex hormone-binding globulin (SHBG) than women with peripheral obesity or normal weight (Pasquali et al., 1990). Reduction of circulating SHBG increases metabolic clearance and production of SHBG-bound steroids, e.g.

* Corresponding author. Address: Charité-Universitätsmedizin Berlin, Campus Benjamin Franklin, Department of Psychiatry, Eschenallee 3, 14050 Berlin, Germany. Tel.: +49 (30) 8445 8796. E-mail address: [email protected] (S. Roepke). 0022-3956/$ - see front matter Ó 2010 Published by Elsevier Ltd. doi:10.1016/j.jpsychires.2010.01.007

testosterone (Kirschner et al., 1990). Also, production rates and metabolic clearance rates of other androgens, such as dehydroandrosterone (DHEA) and androstenedione, are increased in obesity (Pasquali, 2006). Hyperandrogenemia in women is often related to polycystic ovary syndrome (PCOS) Norman, 2002 and PCOS is linked to obesity in women (Gambineri et al., 2002). The exact causality of these associations remains unknown. Thus, some data argue for a primary abdominal fat deposition with hyperinsulinemia and secondary hyperandrogenemia and cyst formation, whereas more recent data suggest a primary androgen excess with secondary visceral fat deposition (Escobar-Morreale and San Millán, 2007). There is also evidence that some of the psychopathological symptoms frequently found in BPD are linked to altered serum androgen levels (Archer, 2006; Hermans et al., 2008). A number of studies, but not all, have demonstrated a positive correlation of serum testosterone (T) and aggressive behavior in animals and, to a lesser degree, in humans within both sexes (Rubinow and Schmidt, 1996; Archer, 2006). Thus, adolescent girls with aggressive conduct disorders showed higher level of FT and lower

848

S. Roepke et al. / Journal of Psychiatric Research 44 (2010) 847–852

level of SHBG compared to healthy controls (Pajer et al., 2006). Among female prisoners, higher levels of T corresponded to increased levels of aggressive dominance and violent behavior (Dabbs and Hargrove, 1997). Finally, endogenous T was higher in bulimic women compared to controls (Sundblad et al., 1994; Cotrufo et al., 2000; Naessén et al., 2006). In one of the latter studies, plasma T correlated positively with aggression (Cotrufo et al., 2000). In summary, these findings suggest a role of elevated T in the symptomatology of disorders with impaired impulse control in women. Nevertheless, the data on aggression and testosterone are conflicting. With respect to the effect of exogenous T, some researchers conclude that an effect of T on human aggression is evident only at pharmacological (i.e., unphysiologically high), rather than at physiological (i.e., replacement), dose levels of the hormone (e.g. Rubinow and Schmidt, 1996). Also, androgen serum levels seem to be associated with mood. Although there are few studies at present, increased and decreased testosterone levels were associated with depressive symptoms in women (Rohr, 2002; Weiner et al., 2004). 2. Objectives of the study In the current study, we examined serum androgen concentrations and PCO-status using ultrasound in women with BPD and healthy controls. Further, the study aimed to examine whether psychometrically measured psychopathology of BPD correlates with serum free testosterone concentrations. Thus, the objective of the study was to evaluate the role of an altered androgen metabolism for diagnosis and symptom profiling in BPD. 3. Materials and methods 3.1. Subjects Forty-one patients with the diagnosis of borderline personality disorder (BPD) according to the DSM-IV fulfilled the inclusion criteria; 31 patients agreed to participate in the study and 27 of them agreed to the pelvic ultrasound examination. As controls, 30 healthy participants were included. All patients were admitted to our specialized inpatient treatment program for BPD during which they were consecutively recruited into the study between October 2005 and March 2007. Prior to admission to the inpatient program, all of them were on a waiting list and none was admitted for acute care. Patients were not reimbursed for study participation and had no advantage or disadvantage by participating in the study. Healthy controls were recruited via media advertisements and reimbursed for participation with 50€. Axis II diagnoses were confirmed or excluded in patients and controls with the German version of the Structured Clinical Interview for DSM-IV (SCID II), axis I diagnoses were assessed with the German version of the Mini International Neuropsychiatric Interview (MINI). Axis I and II interviews were performed by trained psychiatrists or psychologists. Interviewers were not blinded and were aware of the status of the participant (patient or control). The study was approved by the ethics committee of the Charité-Universitätsmedizin Berlin. All participants provided written informed consent after having received a thorough explanation of the study. Exclusion criteria for the patients were anorexia nervosa, oligophrenia, schizophrenia, hormonal contraception within the last six months, pregnancy, alcohol or substance abuse within the last three months, valproate medication within the last three years, and age younger than 18 years. Psychiatric disorders in the control group were also excluded by SCID II and MINI. Sociodemographic parameters of patients and controls are presented in Table 1.

Table 1 Sociodemographic data and psychometric measures in patients with borderline personality disorder and healthy controls.

Age ZAN-BPD sum Affective Cognitive Impulsivity Relations HAMD EDI-2 (bulimia subscale) FAF sum Spontaneous Reactive aggression Irritability Self-aggression Aggression inhibition BIS sum Cognitive Motor Non-planning

Patients (N = 31) Mean ± SD

Controls (N = 30) Mean ± SD

29 ± 6.7 15.1 ± 4.3** 5.4 ± 1.6** 3.6 ± 1.6** 2.5 ± 1.2** 3.6 ± 1.5** 11.2 ± 4.9** 3.3 ± 1.6** 23.6 ± 7.8** 7.0 ± 4.2** 5.8 ± 2.5** 10.7 ± 2.4** 10.0 ± 1.0** 6.6 ± 2.0** 88.4 ± 12.5** 30.3 ± 3.9** 28.9 ± 5.7** 29.2 ± 5.8*

28 ± 4.3 0.9 ± 1.2 0.5 ± 0.9 0.1 ± 0.3 0.1 ± 0.3 0.2 ± 0.6 1.9 ± 1.7 1.5 ± .5 8.5 ± 5.9 2.2 ± 1.9 1.9 ± 2.3 4.4 ± 3.0 2.3 ± 2.0 5.1 ± 1.8 73.8 ± 8.5 24.4 ± 3.9 23.1 ± 3.6 26.3 ± 3.4

ZAN-BPD, Zanarini Rating Scale for Borderline Personality Disorder; HAMD, Hamilton Depression Scale; EDI-2, Eating Disorder Inventory-2; FAF, Questionnaire for Measuring Factors of Aggression; BIS, Barratt Impulsiveness Scale. * p < 0.05. ** p < 0.01.

About 19 of the 31 patients had recurrent Major Depressive Disorder (MDD) and 13 fulfilled the criteria of actual MDD. Ten patients fulfilled criteria for dysthymia, nine for agoraphobia or panic disorder, nine for social phobia, three for obsessive–compulsive disorder, ten for alcohol abuse within the past 12 months, 12 for substance abuse within the past 12 months, and 14 for bulimia. During the study, 16 patients received SSRI, mainly for the treatment of MDD, affective symptoms in BPD or impulsivity. Ten patients received atypical neuroleptics (aNL) for the treatment of impulsivity and affective symptoms. Nine patients were without medication and five patients received aNL and SSRI. The patients did not receive further psychotropic concomitant medication. Two patients with BPD reported valproate intake in the past (more than three years ago), in both cases for less than four months. No patient had documented valproate intake in the medical records. No women from the control group received any medication within the last three months. 3.2. Psychometric instruments Borderline pathology was assessed with the Zanarini Rating Scale for Borderline Personality Disorder (ZAN-BPD), a clinicianadministered scale for the assessment of change in DSM-IV borderline psychopathology. The scale measures psychopathology over the last week. Each of the nine criteria for BPD is rated on a fivepoint anchored rating scale from 0 to 4, yielding a total score of 0–36 (Zanarini et al., 2003). Depressive symptoms were assessed with the 17-item Hamilton Depression Scale (HAMD). Impulsivity was assessed with the Barratt Impulsiveness Scale (BIS) version 10 Barratt, 1985, measuring motor, non-planning and cognitive components of impulsivity on a four-point anchored rating scale from 1 to 4. The Questionnaire for Measuring Factors of Aggression (FAF) Hampel and Selg, 1975, a German adaptation of the Buss– Durkee Hostility Inventory (Buss and Durkee, 1957), was used to evaluate various components of aggressive behavior, such as spontaneous and reactive aggression, irritability, self-aggression, openness, and aggression inhibition. Bulimic symptoms were assessed with the bulimia subscale of the Eating Disorder Inventory-2 (EDI-2) Garner, 1991.

849

S. Roepke et al. / Journal of Psychiatric Research 44 (2010) 847–852

3.3. Study design Pelvic ultrasound examination was performed with a 5–9 MHz transvaginal transducer with color and pulsed Doppler facilities (GE, Voluson 730 Expert). Ovarian volume was determined by the following formula: 0.523  length  breadth  width. Follicle numbers were assessed in both longitudinal and anteroposterior cross-sections of the ovary. PCO was determined by the presence of 10 or more subcapsular follicular cysts, 2–8 mm in diameter, arranged around or within thickened ovarian stroma in at least one ovary. Ovary volume was calculated as mean of the volume of both ovaries. Conclusive transvaginal pelvic ultrasound was performed in 23 patients with BPD (74.2% of the BPD group) and 29 healthy controls (96.7% of the control group). Eight patients with BPD had no conclusive examination, four patients refused the examination, three results were not conclusive due to masking abdominal air, and one examination was not possible due to technical problems. One control person refused the pelvic ultrasound. 3.4. Hormone assays Prolactine, LH, FSH, estradiol, progesterone, SHBG, and cortisol were assessed with a fluoroimmunoassay (Perkin–Elmer Life Science Products, Inc., Boston, MA). DHEAS, 17-OHP, and A were assessed with a radio immunoassay (iodine-125) kit (Diagnostic Systems Laboratories, Sinsheim, Germany). Testosterone and estrone were assessed with an immunoradiometric assay using antiserum from Biogenesis/Quartett (Berlin, Germany) and a tracer (H3) from New England Nuclear (NEN)/Perkin–Elmer (Life Science Products, Inc., Boston, MA). Free androgen index was calculated by the following formula: (T/SHBG)  100. Two fasting blood samples were drawn between the third and fifth day of the menstrual cycle in the morning between 8 and 9 a.m. with a 30 min interval. Patients were in a lying position 30 min before the first and until the second blood sample was taken. One hour after samples were taken, the blood was centrifuged. Equal amounts of serum from each of the two samples were then mixed and send for lab testing. Glucose was tested prior to blood sampling to verify fasting. 3.5. Statistics Statistical analysis was performed using the SPSS (version 14) software. Before use of parametric tests, histograms and Kolmogo-

rov–Smirnov-test were performed. Between groups comparisons were done with two tailed t-tests. The distribution of ultrasonographical PCO in the two study groups was assessed with cross tables and analysed with a chi-square test. Further analysis was performed with multiple analysis of covariance (MANCOVA), and Pearson coefficients of correlation were calculated. All significance levels were set to 0.05. All values were given as mean and standard deviation when appropriate. 4. Results 4.1. Prevalence of PCO About 7 out of 24 patients with BPD had PCO (30.4%), four of them on both ovaries. Two healthy controls out of 29 had PCO (6.9%), both bilaterally. The prevalence of PCO was significantly higher in BPD patients compared to the control group (v2(1) = 4.97, p = 0.03). Mean ovarian volume of both ovaries was not significantly different between the PCO and non-PCO group (data not shown). The prevalence of Axis I diagnosis of eating disorder, recent or lifetime MDD and post-traumatic stress disorder (PTSD) were not significantly different between BPD patients with or without PCO (data not shown). Only dysthymia was more prevalent in the BPD with PCO group with a trend towards significance (v2(1) = 3.74, p = 0.053). The higher BMI in BPD patients with PCO (27.4 ± 7.5) compared to BPD patients without PCO (25.4 ± 8.3) was not significant (t = 0.6, df = 22, p = 0.59). 4.2. Androgen serum level A t-test revealed significant differences between BPD patients and controls with regard to the BMI (t = 3.6, df = 59, p = 0.01), FT (t = 2.15, df = 58, p = 0.04), androstenedione (t = 2.4, df = 58, p = 0.02), 17-OHP (t = 2.5, df = 58, p = 0.02), and prolactine (t = 2.2, df = 58, p = 0.03) (Table 2). A MANCOVA with the hormones as the dependent variable and the covariates BMI, SSRI medication, and atypical neuroleptic medication revealed significant differences between the groups (Wilks-Lambda, F = 3.40, df = 32, p = 0.02, partial g2 = 0.58), for the covariate BMI (Wilks-Lambda, F = 2.88, df = 32, p = 0.007, partial g2 = 0.53), and also revealed a trend for the interaction of PCO-status and group (Wilks-Lambda, F = 1.87, df = 32, p = 0.074, partial g2 = 0.43). The influence of SSRI

Table 2 Hormone measures in patients with borderline personality disorder and healthy controls. Patients (N = 31) Mean ± SD

BMI FT (T/SHBG) Testosterone (pg/ml) SHBG (nmol/l) Androstenedione (ng/ml) 17-OHP (pg/ml) Progesterone (ng/ml) DHEAS (lg/ml) FSH (mIU/ml) LH (mIU/ml) Prolactin (lIU/ml) Estradiol (pg/ml) Estrone (pg/ml) Cortisol (ng/ml)

Controls (N = 30) Mean ± SD

All (N = 31)

PCO (N = 7)

Non-PCO (N = 16)

All (N = 30)

PCO (N = 2)

Non-PCO (N = 27)

26.1 ± 7.4* 764.0 ± 402.0* 337.4 ± 99.2 53.4 ± 25.0 1.7 ± 0.7* 979.0 ± 423.7* 0.5 ± 0.2 1.6 ± 0.9 5.6 ± 1.6 4.8 ± 1.7 413.2 ± 312.9* 33.8 ± 16.8 54.0 ± 19.8 113.7 ± 31.4

27.42 ± 7.48 861.9 ± 561.9 364 ± 85.3 54.7 ± 27.7 2.04 ± .74 1353 ± 489 0.60 ± .24 1.31 ± .59 6.3 ± 1.9 5.0 ± 1.4 572 ± 423 33.3 ± 12.8 51.9 ± 16.6 127.4 ± 19.6

25.69 ± 8.50 778.1 ± 351.1 349 ± 104.1 51.5 ± 21.4 1.62 ± .79 840 ± 347 0.43 ± .12 1.55 ± 1.01 5.5 ± 1.4 4.8 ± 2.0 333 ± 176 34.8 ± 19.4 60.1 ± 21.8 108.6 ± 34.1

21.1 ± 1.6 568.9 ± 285.5 303.1 ± 87.5 60.9 ± 22.8 1.3 ± 0.3 741.7 ± 308.8 0.50 ± 0.2 1.4 ± 0.5 5.4 ± 1.4 4.1 ± 1.7 273.2 ± 148.7 32.9 ± 10.8 47.2 ± 12.2 111.5 ± 26.6

20.23 ± 1.52 411.4 ± 151.3 287.5 ± 113.8 69.5 ± 2.1 1.19 ± .45 531.0 ± 149.9 0.45 ± .21 1.09 ± .16 4.9 ± .2 3.8 ± .9 208.0 ± 70.7 35.5 ± 13.4 49.5 ± 5.0 126.5 ± 13.4

21.16 ± 1.59 581.5 ± 297.0 303.8 ± 89.6 60.4 ± 24.0 1.33 ± .31 729.1 ± 282.2 0.50 ± .24 1.42 ± .47 5.5 ± 1.4 4.1 ± 1.8 264.0 ± 136.5 32.6 ± 11.1 47.0 ± 12.9 109.6 ± 27.4

FT: calculated free testosterone (T/SHBG  100), SHBG: sex steroid hormone-binding globulin, 17-OHP: 17a-hydroxyprogesterone, DHEAS: dehydroandrosterone sulphate, FSH: follicle stimulating hormone, LH: luteinizing hormone. * p < 0.05 (t-test between groups patients/controls).

850

S. Roepke et al. / Journal of Psychiatric Research 44 (2010) 847–852

medication (Wilks-Lambda, F = 2.04, df = 32, p = 0.051, partial g2 = 0.45) and medication with atypical neuroleptics (Wilks-Lambda, F = 1.77, df = 32, p = 0.094, partial g2 = 0.41) revealed trends toward significance. The PCO-status alone did not have a significant effect on hormone levels (F = 0.56). Including ovarian volume as a covariate did not change results significantly (data not shown). Univariate tests showed that group membership was significantly related to testosterone (F = 4.56, df = 1, p = 0.038, partial g2 = 0.09), FT (F = 9.46, df = 1, p = 0.004, partial g2 = 0.17), 17-OHP (F = 21.71, df = 1, p < 0.001, partial g2 = 0.33), androstenedione (F = 14.96, df = 1, p < 0.001, partial g2 = 0.25), and cortisol (F = 7.28, df = 1, p = 0.01, partial g2 = 0.14). Univariate tests showed also that BMI was significantly related to FT (F = 4.23, df = 1, p = 0.046, partial g2 = 0.08), androstenedione (F = 5.12, df = 1, p = 0.029, partial g2 = 0.10), cortisol (F = 5.64, df = 1, p = 0.022, partial g2 = 0.11), SHBG (F = 5.44, df = 1, p = 0.024, partial g2 = 0.11), and DHEAS (F = 4.57, df = 1, p = 0.038, partial g2 = 0.09). Only the correlations between BMI and SHBG (r = 0.39, p = 0.002) and between BMI and cortisol (r = 0.26, p = 0.048) proved to be significant within the whole sample, whereas the BMI correlation was not significant with 17-OHP (r = 0.08, p = 0.95), androstenedione (r = 0.14, p = 0.28), and DHEAS (r = 0.15, p = 0.25). In the BPD group, all androgens were positively correlated with their respective precursor proteins: FT with T (r = 0.40, p = 0.03), T with A (r = 0.69, p < 0.01), A with DHEAS (r = 0.47, p < 0.01), and A with 17-OHP (r = 0.73, p < 0.01). Only 17-OHP was not significantly correlated with progesterone (r = 0.22, p = 0.24). 4.3. Influence of androgens and BMI on psychopathology Standard multiple linear regression models with impulsivity (BIS sum score), bulimic symptoms (EDI-2, bulimia subscale), aggression (FAF sum score) and depressive symptoms (HAMD) as dependent variables and FT, BMI and interaction (BMI  FT) as independent variables revealed a significant model only for depressive symptoms (r2 = 0.28, p = 0.031). Thus, depressive symptoms were significantly associated with BMI, FT and interaction (Table 3). There was a significant positive correlation between HAMD and FT (r = 0.36, p = 0.045), that lost significance after we controlled for BMI (r = 0.34, p = 0.068). 5. Discussion The major finding of the present study is the increased prevalence of polycystic ovaries in patients with borderline personality disorder (30.4%) compared to healthy controls (6.9%). Furthermore, patients with BPD had significantly increased serum androgen concentrations of testosterone, FT, androstenedione, and 17-OHP compared to the control group, also independently of the BMI. By contrast, PCO-status as defined by ultrasonography did not significantly influence serum androgen concentrations. This might indicate a hyperandrogenic state in BPD independent of ovarian cysts. To interpret these data we have to consider the vicious circle of the pathogenesis of polycystic ovary syndrome (PCOS), which is

still not fully understood (Escobar-Morreale and San Millán, 2007). Disturbed glucose metabolism and compensatory hyperinsulinemia can stimulate follicle formation via increased sensitivity of granulosa cells to FSH (Fulghesu et al., 1997). Disturbed glucose metabolism has been found in BPD patients (Kahl et al., 2005). Furthermore, elevated ovarian insulin concentration facilitates intraovarian hyperandrogenism, which contributes to the arrest of follicular maturation characteristic of PCOS patients (Jonard and Dewailly, 2004). Theca cells propagated in long-term culture from patients with full PCOS have increased androgen biosynthesis which persists after three to four passages in culture, which argues against an influence of the paracrine or endocrine in vivo milieu to stimulate androgen production (Wickenheisser et al., 2006). Thus, ovarian cysts are an autonomic source of elevated androgens, at least in later stages of the disease. Also, increased abdominal fat, which has also been demonstrated in BPD patients (Kahl et al., 2005), contributes independently to hyperandrogenemia via hyperinsulinism, hypoadiponectinemia (Escobar-Morreale et al., 2006), cytokine excess (Fernandez-Real, 2003), and increased oxidative stress (González et al., 2006). Hyperinsulinemia (Poretsky et al., 1999; Dunaif, 1999) also inhibits hepatic SHBG synthesis (Plymate et al., 1988) and thus increases FT. Our finding of an inverse correlation between BMI and SHBG is in line with prior results (Plymate et al., 1988; Pugeat et al., 1991). Also, a significant BMI influence on FT, DHEAS, and androstenedione in our sample was found. Recent research data indicate androgen excess to be a contributing factor in abdominal fat deposition (Elbers et al., 2003) both during fetal period (Abbott, 2007) and infancy (Franks, 2002). Modified androgen metabolism could be genetically determined (Franks et al., 2000). Our findings of significantly increased testosterone, FT, androstenedione, and 17-OHP in the patient group, independently of the increased BMI, could indicate a disturbed androgen metabolism in BPD. In summary, several pathophysiological mechanisms are candidates for the origin of the vicious circle in PCOS: increased insulin resistance, increased abdominal fat, increased or disturbed androgen metabolism. Disturbed or normal glucose metabolism and the presence or absence of cysts has been found in recent studies on PCOS patients. These partially conflicting results could either present different stages of one pathophysiological mechanism or different interacting disorders (Escobar-Morreale and San Millán, 2007). This may explain our finding of a PCO-independent androgen elevation. Naessén et al. (2006) found a higher prevalence of PCOS in bulimic women. Bulimic patients had more menstrual disturbances and elevated T/SHBG ratios while the prevalence of PCO on ultrasonography was not significantly different between the groups. In our BPD sample, there was no association between bulimia and PCO-status which is in line with these findings. As in our sample, Naessén et al. (2006) did not find any correlation between endocrine and clinical variables of PCOS. Moreover, we found significantly higher FT in depressed patients independently of BMI. Also, a positive correlation between depressive symptoms and FT in the BPD sample could be demonstrated, although the correlation was only seen on a trend-level

Table 3 Association between psychometric measures and free testosterone and BMI in patients with borderline personality disorder. BIS sum score

FT BMI Interaction

FAF sum score

EDI-2, bulimia subscale

HAMD

ba

p

ba

p

ba

p

ba

p

0.24 0.49 0.60

0.76 0.31 0.55

0.11 0.34 0.12

0.89 0.49 0.91

0.47 0.23 0.17

0.51 0.62 0.86

1.81 0.90 1.95

0.010 0.039 0.029

BIS, Barratt Impulsiveness Scale; FAF, Questionnaire for Measuring Factors of Aggression; EDI-2, Eating Disorder Inventory-2; HAMD, Hamilton Depression Scale; FT, free testosterone; BMI, body–mass-index; interaction, BMI  FT. a Standardized regression coefficient from multiple regression procedure.

S. Roepke et al. / Journal of Psychiatric Research 44 (2010) 847–852

when we controlled for BMI. These results reflect recent data demonstrating increased state and trait depressive symptoms in PCOS women (Weiner et al., 2004). Weiner et al. (2004) also found a positive correlation between FT and depressive symptoms within the normal range of FT in their BMI matched samples. In our sample, we did not find a correlation between impulsivity and serum androgen concentrations. These results are in favour of the hypothesis that the relationship between basal testosterone concentrations and aggression is only modest at best e.g. (Archer, 1991). Our results are of special interest, as obesity in BPD is a negative predictor for remission from the disorder (Frankenburg and Zanarini, 2004) and associated with different medical conditions (diabetes, hypertension, osteoarthritis, chronic back pain, carpal tunnel syndrome, urinary incontinence, gastroesophageal reflux disorder, gallstones, and asthma) (Frankenburg and Zanarini, 2006). Moreover, PCOS is associated with an increased risk of miscarriage, hyperlipidemia, cardiovascular disease, endometrial cancer and type 2 diabetes mellitus (Pasquali and Gambineri, 2006). We suggest that replication studies in larger samples are needed since implications for the physical health of affected patients are far reaching. A confirmation of our study results would call for PCOS screening in obese BPD patients, as efficient non-pharmacological (Norman et al., 2002) and pharmacological (De Leo et al., 2003; Jayagopal et al., 2005; Sabuncu et al., 2003) treatments exist. 5.1. Limitations Our results are limited by the sample size. Furthermore, due to the lack of precision of ratings of clinical PCOS features (e.g. hirsutism) with inexperienced raters, we exclusively assessed ultrasonographic PCO and androgen serum concentrations in our study, and not full PCOS criteria. Thus, our interpretation is restricted to PCO with elevated androgens, not PCOS. Many women with PCO do not have PCOS as defined by the NIH (Polson et al., 1988). Nevertheless, the prevalence of PCO in our healthy sample is comparable to the 4–7% prevalence of PCOS in the general population of reproductive-aged women (Knochenhauer et al., 1998). Two patients in our sample reported prior valproate medication and all patients had multiple hospitalisations and multiple psychotropic medications in the past. As the pharmacological agent valproate, commonly used in BPD treatment, is associated with PCOS (Rasgon, 2004), we cannot fully exclude this confounding factor. Studies of bipolar disorder and epilepsy suggest that at least some features of PCOS may precede valproate treatment (Rasgon, 2004; Rasgon et al., 2005a,b). Valproate could act as an enhancer of PCOS, possibly via weight gain. The interaction effect of PCO-status and group indicating significantly higher 17-OHP in the PCO control group compared to the non-PCO control group is possibly due to the low number of subjects in the PCO-positive control group. The influence of neuroleptic medication and SSRIs on androgen metabolism seen as a trend in our model should be addressed in future studies. Also, previous research has indicated a possible influence of SSRI medication on metabolic syndrome and glucose metabolism (Raeder et al., 2006). Thus, findings of disturbed glucose metabolism in BPD need to be re-evaluated in light of the high prevalence of SSRI medication in this study population (Kahl et al., 2005). Elevated androgens, obesity, and PCO are not specific for BPD since depression has been linked to increased visceral fat (Kahl et al., 2005) and disturbed glucose metabolism (Rasgon et al., 2005b). This is particularly important since dysthymia appeared to be slightly more prevalent in our PCO-positive BPD subsample. Thus, future studies should be designed to include a clinical control group of age and BMI matched MDD patients.

851

Nevertheless, Kahl et al. (2005) found higher visceral fat in the MDD group with comorbid BPD compared to the MDD group without comorbid BPD and disturbed glucose metabolism was only significantly increased in the comorbid group. Thus, MDD seems to be an additional contributing factor to increased visceral fat and disturbed glucose metabolism. In summary, our preliminary results suggest that BPD is associated with a significantly higher incidence of PCO and significantly elevated androgen serum concentrations. FT seems to have a significant influence on state depressive symptoms, but not on impulsivity or bulimic symptoms. Role of the funding source This research was supported in part by a Grand from CharitéUniversity Medicine Berlin. The funding sources had no role in this study’s design, in the collection, analysis and interpretation of data; in the writing of the report; and in the decision to submit the paper for publication. Contributors Stefan Roepke: Dr. Roepke designed and conducted the study, undertook statistical analyses and wrote the manuscript. Andreas Ziegenhorn: Dr. Ziegenhorn undertook statistical analyses and wrote the manuscript. Julia Kronsbein: Ms. Kronsbein conducted the study and assisted in drafting the manuscript. Angela Merkl: Dr. Merkl collected the data and assisted in drafting the manuscript. Scharif Bahri: Mr. Bahri collected the data and assisted in drafting the manuscript. Julia Lange: Dr. Lange conducted the study and assisted in drafting the manuscript. Horst Lübbert: Dr. Lübbert designed the study and assisted in drafting the manuscript. Ulrich Schweiger: Dr. Schweiger undertook statistical analyses and wrote the manuscript. Isabella Heuser: Dr. Heuser undertook statistical analyses and assisted in drafting the manuscript. Claas-H. Lammers: Dr. Lammers designed the study and assisted in drafting the manuscript. Conflict of interests Dr. Stefan Roepke has received research grants from Lilly and AstraZeneca. He has received teaching honoraries from Wyeth and AstraZeneca. Dr. Isabellla Heuser has received research grants from Jansen Cilag and Merz. She has advised Bayer Schering and has received teaching honoraries from Wyeth and Pfizer. Drs. Andreas Ziegenhorn, Julia Kronsbein, Angela Merkl, Scharif Bahri, Julia Lange, Horst Lübbert, Ulrich Schweiger, and Claas-H. Lammers report no conflict of interests. Acknowledgments We thank Vincent Ballenas, M.Sc, Montreal, for assistance with proofreading and for many helpful comments. Further, we thank Heinrich Wernze for fundamental help with the study design. References Abbott DH. Fetal programming of polycystic ovary syndrome. In: Kovacs G, Norman R, editors. Polycystic ovary syndrome. Cambridge: University Press; 2007. p. 262–87. Archer J. The influence of testosterone on human aggression. British Journal of Psychology 1991;82:1–28. Archer J. Testosterone and human aggression: an evaluation of the challenge hypothesis. Neuroscience & Biobehavioral Reviews 2006;30:319–45. Barratt ES. Impulsiveness substrates: arousal and information processing. In: Spence JT, Izard CE, editors. Motivation, emotion and personality. NorthHolland: Elsevier Science; 1985. p. 137–46. Buss AH, Durkee A. An inventory for assessing different kinds of hostility. Journal of Consulting Psychology 1957;21:343–9.

852

S. Roepke et al. / Journal of Psychiatric Research 44 (2010) 847–852

Cotrufo P, Monteleone P, d’Istria M, Fuschino A, Serino I, Maj M. Aggressive behavioral characteristics and endogenous hormones in women with bulimia nervosa. Neuropsychobiology 2000;42:58–61. Dabbs Jr JM, Hargrove MF. Age, testosterone, and behavior among female prison inmates. Psychosomatic Medicine 1997;59:477–80. De Leo V, la Marca A, Petraglia F. Insulin-lowering agents in the management of polycystic ovary syndrome. Endocrine Reviews 2003;24:633–67. Dunaif A. Insulin action in the polycystic ovary syndrome. Endocrinology Metabolism Clinics of North America 1999;28:341–59. Elbers JM, Giltay EJ, Teerlink T, Scheffer PG, Asscheman H, Seidell JC, et al. Effects of sex steroids on components of the insulin resistance syndrome in transsexual subjects. Clinical Endocrinology 2003;58:562–71. Escobar-Morreale HF, San Millán JL. Abdominal adiposity and the polycystic ovary syndrome. Trends in Endocrinology and Metabolism 2007;18:266–72. Escobar-Morreale HF, Villuendas G, Botella-Carretero JI, Alvarez-Blasco F, Sanchón R, Luque-Ramírez M, et al. Adiponectin and resistin in PCOS: a clinical, biochemical and molecular genetic study. Human Reproduction 2006;21: 2257–65. Fernandez-Real JM, Ricart W. Insulin resistance and chronic cardiovascular inflammatory syndrome. Endocrine Reviews 2003;24:278–301. Frankenburg FR, Zanarini MC. The association between borderline personality disorder and chronic medical illnesses, poor health-related lifestyle choices, and costly forms of health care utilization. Journal of Clinical Psychiatry 2004;65:1660–5. Frankenburg FR, Zanarini MC. Personality disorders and medical comorbidity. Current Opinion in Psychiatry 2006;19:428–31. Franks S. Adult polycystic ovary syndrome begins in childhood. Best Practice & Research Clinical Endocrinology & Metabolism 2002;16:263–72. Franks S, Gilling-Smith C, Gharani N, McCarthy M. Pathogenesis of polycystic ovary syndrome: evidence for a genetically determined disorder of ovarian androgen production. Human Fertility 2000;3:77–9. Fulghesu AM, Villa P, Pavone V, Guido M, Apa R, Caruso A, et al. The impact of insulin secretion on the ovarian response to exogenous gonadotropins in polycystic ovary syndrome. Journal of Clinical Endocrinology & Metabolism 1997;82:644–8. Gambineri A, Pelusi C, Vicennati V, Pagotto U, Pasquali R. Obesity and the polycystic ovary syndrome. Journal of Clinical Endocrinology & Metabolism 2002;26: 883–96. Garner DM. Eating Disorder Inventory-2 manual. Odessa, Florida: Psychological Assessment Resources; 1991. González F, Rote NS, Minium J, Kirwan JP. Reactive oxygen species-induced oxidative stress in the development of insulin resistance and hyperandrogenism in polycystic ovary syndrome. Journal of Clinical Endocrinology & Metabolism 2006;91:336–40. Hampel R, Selg H. FAF – questionnaire for measuring factors of aggression. Göttingen: Hogrefe; 1975. Hermans EJ, Ramsey NF, van Honk J. Exogenous testosterone enhances responsiveness to social threat in the neural circuitry of social aggression in humans. Biological Psychiatry 2008;63:263–70. Jayagopal V, Kilpatrick ES, Holding S, Jennings PE, Atkin SL. Orlistat is as beneficial as metformin in the treatment of polycystic ovarian syndrome. Journal of Clinical Endocrinology & Metabolism 2005;90:729–33. Jonard S, Dewailly D. The follicular excess in polycystic ovaries, due to intra-ovarian hyperandrogenism, may be the main culprit for the follicular arrest. Human Reproduction Update 2004;10:107–17. Kahl KG, Bester M, Greggersen W, Rudolf S, Dibbelt L, Stoeckelhuber BM, et al. Visceral fat deposition and insulin sensitivity in depressed women with and without comorbid borderline personality disorder. Psychosomatic Medicine 2005;67:407–12. Kahl KG, Bens S, Ziegler K, Rudolf S, Dibbelt L, Kordon A, et al. Cortisol, the cortisol– dehydroepiandrosterone ratio, and pro-inflammatory cytokines in patients with current major depressive disorder comorbid with borderline personality disorder. Biological Psychiatry 2006;59:667–71. Kirschner MA, Samojlik E, Drejka M, Szmal E, Schneider G, Ertel N. Androgen– estrogen metabolism in women with upper body versus lower body obesity. Journal of Clinical Endocrinology & Metabolism 1990;70:473–9.

Knochenhauer ES, Key TJ, Kahsar-Miller M, Waggoner W, Boots LR, Azziz R. Prevalence of the polycystic ovary syndrome in unselected black and white women of the southeastern United States: a prospective study. Journal of Clinical Endocrinology & Metabolism 1998;83:3078–82. Naessén S, Carlström K, Garoff L, Glant R, Hirschberg AL. Polycystic ovary syndrome in bulimic women – an evaluation based on the new diagnostic criteria. Gynecological Endocrinology 2006;22:388–94. Norman RJ. Hyperandrogenaemia and the ovary. Molecular and Cellular Endocrinology 2002;191:113–9. Norman RJ, Davies MJ, Lord J, Moran LJ. The role of lifestyle modification in polycystic ovary syndrome. Trends in Endocrinology and Metabolism 2002;13:251–7. Pajer K, Tabbah R, Gardner W, Rubin RT, Czambel RK, Wang Y. Adrenal androgen and gonadal hormone levels in adolescent girls with conduct disorder. Psychoneuroendocrinology 2006;31:1245–56. Pasquali R. Obesity and androgens: facts and perspectives. Fertility and Sterility 2006;85:1319–40. Pasquali R, Gambineri A. Polycystic ovary syndrome: a multifaceted disease from adolescence to adult age. Annals of the New York Academy of Sciences 2006;1092:158–74. Pasquali R, Casimirri F, Platè L, Capelli M. Characterization of obese women with reduced sex hormone-binding globulin concentrations. Hormone and Metabolic Research 1990;22:303–6. Plymate SR, Matej LA, Jones RE, Friedl KE. Inhibition of sex hormone-binding globulin production in the human hepatoma (Hep G2) cell line by insulin and prolactin. Journal of Clinical Endocrinology & Metabolism 1988;67:460–4. Polson DW, Adams J, Wadsworth J, Franks S. Polycystic ovaries – a common finding in normal women. Lancet 1988;8590:870–2. Poretsky L, Cataldo NA, Rosenwaks Z, Giudice LC. The insulin-related ovarian regulatory system in health and disease. Endocrine Reviews 1999;20: 535–82. Pugeat M, Crave JC, Elmidani M, Lejeune H, Charrié A, Fleury MC, et al. Inverse relationship between body mass index and fasting insulinemia with testosterone-binding protein in hirsutism. Annales d’endocrinologie 1991;52:93–6. Raeder MB, Bjelland I, Emil Vollset S, Steen VM. Obesity, dyslipidemia, and diabetes with selective serotonin reuptake inhibitors: the Hordaland Health Study. Journal of Clinical Psychiatry 2006;67:1974–82. Rasgon N. The relationship between polycystic ovary syndrome and antiepileptic drugs: a review of the evidence. Journal of Clinical Psychopharmacology 2004;24:322–34. Rasgon NL, Altshuler LL, Fairbanks L, Elman S, Bitran J, Labarca R, et al. Reproductive function and risk for PCOS in women treated for bipolar disorder. Bipolar Disorders 2005a;7:246–59. Rasgon NL, Reynolds MF, Elman S, Saad M, Frye MA, Bauer M, et al. Longitudinal evaluation of reproductive function in women treated for bipolar disorder. Journal of Affective Disorders 2005b;89:217–25. Rohr UD. The impact of testosterone imbalance on depression and women’s health. Maturitas 2002;41(Suppl. 1):25–46. Rubinow DR, Schmidt PJ. Androgens, brain, and behaviour. The American Journal of Psychiatry 1996;153:974–84. Sabuncu T, Harma M, Harma M, Nazligul Y, Kilic F. Sibutramine has a positive effect on clinical and metabolic parameters in obese patients with polycystic ovary syndrome. Fertility and Sterility 2003;80:1199–204. Sundblad C, Bergman L, Eriksson E. High levels of free testosterone in women with bulimia nervosa. Acta Psychiatrica Scandinavica 1994;90:397–8. Weiner CL, Primeau M, Ehrmann DA. Androgens and mood dysfunction in women: comparison of women with polycystic ovarian syndrome to healthy controls. Psychosomatic Medicine 2004;66:356–62. Wickenheisser JK, Nelson-DeGrave VL, McAllister JM. Human ovarian theca cells in culture. Trends in Endocrinology and Metabolism 2006;17:65–71. Zanarini MC, Vujanovic AA, Parachini EA, Boulanger JL, Frankenburg FR, Hennen J. Zanarini Rating Scale for Borderline Personality Disorder (ZAN-BPD): a continuous measure of DSM-IV borderline psychopathology. Journal of Personality Disorders 2003;17:233–42.