Salivary testosterone is associated with feelings of senselessness and self-dislike in women with borderline personality disorder Authors: Eugenia Kulakova, Livia Graumann, An Bin Cho, Christian Eric Deuter, Oliver T. Wolf, Julian Hellmann-Regen, Stefan Roepke, Christian Otte, Katja Wingenfeld Published: 2024 ABSTRACT ABSTRACTBackground: Women with borderline personality disorder (BPD) show increased basal levels of testosterone. We investigated whether salivary testosterone levels in women with BPD were indicative of specific symptoms associated with BPD. Based on the assumed link between testosterone and interpersonal dominance, we hypothesized a positive association between testosterone and externalising, i.e. aggressive or impulsive behaviour, potentially contributing to higher burden of interpersonal reactivity and conflict.Methods: Saliva was collected from 98 women with BPD (average age in years: 28, range 18–46) between 1 and 2 pm. Self-rating scales were administered to assess severity of BPD (Borderline Symptom Checklist, BSL-23) and depressive symptoms (Beck’s Depression Inventory, BDI-II). Regression analyses targeted associations between individual testosterone levels and BSL-23 and BDI-II total and by-item scores.Results: Higher testosterone levels were associated with higher overall disease burden indicated by BSL-23 and BDI-II total scores. When analysed by item, higher testosterone levels were significantly associated with increased feelings of self-dislike, senselessness and pessimism, and the feeling of being a failure.Conclusion: Our findings show that in women with BPD testosterone levels are positively associated with increased borderline and depressive symptomatology. Contrary to our expectations, rather than predicting externalising symptoms, higher testosterone is associated with a well-defined cluster of internalising symptoms characterized by a pessimistic and derogatory view towards oneself. Introduction Borderline Personality Disorder (BPD) is a debilitating mental illness that affects 1–2% of the general population. Central symptoms include instability of affect, identity and social interactions. Arising emotional tension can be overwhelming and lead to aggressive, self-harming or suicidal behaviour, as bottom-up intense affect interacts with dysfunctional negative beliefs about oneself and others. Multiple factors contribute to BPD aetiology, with early life adversity and dysregulation of the stress response interacting with neuro-endocrine development and functioning throughout life (Eck & Bangasser, 2020; Gunderson et al., 2018). In contrast to stress hormones, the interplay of gonadal hormones and BPD symptomatology has only recently started to gain attention. Female patients with BPD show increased testosterone levels in saliva (Dyson et al., 2023; Rausch et al., 2015), serum (Bonfig et al., 2022; Roepke et al., 2010) and hair (Dettenborn et al., 2016), suggesting a disturbed androgen metabolism with higher incidence of hyperandrogenism (Tan et al., 2018). In premenopausal women, one quarter of testosterone is produced in the adrenal glands (Longcope, 1986). As such, deviant corticoid functioning and a chronically active adrenal system in BPD might lead to increased testosterone secretion, especially in women. Accordingly, baseline testosterone levels show a positive association with perceived life stress in healthy women (King et al., 2005). Historically considered as the ‘male’ hormone, testosterone was associated with stereotypically male behaviour such as aggression, competition and dominance (Crespi, 2016; Eisenegger et al., 2011). Testosterone levels in females are 10% of those found in males. But also in women, baseline testosterone levels show positive correlations with aggression (Harris et al., 1996; Von der Pahlen et al., 2002), dominance (Grant & France, 2001) and competition (Cashdan, 2003), and testosterone supplementation can increase hostile behaviour in women (Sherwin & Gelfand 1985). However, data on the effects of gonadal hormones on premenopausal women is still scarce and the effects of testosterone on psychiatric symptomatology are poorly understood. Interestingly, in a sample of female BPD patients, baseline salivary testosterone did not predict aggressive traits during adolescence (Cavelti et al., 2022) or motherhood (Rausch et al., 2015). Instead, testosterone levels were correlated with depressive and borderline symptoms (Roepke et al., 2010). Similar associations with depressive and anxiety symptoms were observed in adolescents with polycystic ovary syndrome (PCOS) (Donbaloğlu et al., 2022), a gynaecological condition linked to hyperandrogenism and a frequent comorbidity of BPD (Roepke et al., 2010). Notably, in MDD patients without BPD this association is usually reversed, with lower testosterone levels indicative of increased depressive symptoms (Flores-Ramos et al., 2019; Kumsar et al., 2014). Taken together, these findings raise the question which specific depressive and borderline symptoms are increased as a function of higher testosterone in female BPD patients. Aim of the present study In the present study, we investigated whether salivary testosterone in women with BPD was associated with increased symptom burden. Clinical symptoms were assessed with self-rating scales of borderline (BSL-23) and depressive symptoms (BDI-II), as well as clinician-rated presence of DSM-V criteria of BPD (SCID). We were particularly interested in identifying specific symptoms or symptom clusters associated with testosterone. Therefore, in addition to analysing questionnaire sum scores, by-item analyses were conducted. We hypothesized an association of higher testosterone levels with items indicative of externalising, i.e. aggressive or impulsive behaviour. Methods Participants The sample consisted of 98 women (female sex assigned at birth assessed through self-report) with BPD which was a subsample of a larger study (Graumann et al., 2023). Diagnostic criteria for BPD were assessed with the Structured Clinical Interview for DSM-V (SCID) (Beesdo-Baum et al., 2019), exclusion criteria were acute major depressive episode (moderate or severe), current substance abuse, psychotic disorder and intake of more than three different psychotropic substances or benzodiazepines. All participants provided written informed consent and received monetary compensation (minimum 60€) for their participation as part of a larger study. The study was approved by the local ethical committee and in accordance with the Declaration of Helsinki. Procedure Participants underwent diagnostic interviews and filled out questionnaires on psychopathological symptoms using the web-based application REDCap on a tablet or computer, including the Beck Depression Inventory (BDI-II) and the Borderline Symptom List short version (BSL-23). Both scales are well established in clinical and research practice and offer good psychometric properties (Beck et al., 1996; Bohus et al., 2009; Kühner et al., 2007). The BDI-II is a 21-item, self-report rating that measures characteristic attitudes and symptoms of depression within the last 14 days (Beck et al., 1996). The 21 items include (1) Sadness, (2) Pessimism, (3) Past failure, (4) Loss of pleasure, (5) Guilt feelings, (6) Punishment feelings, (7) Self-dislike, (8) Self-criticalness, (9) Suicidal thoughts or wishes, (10) Crying, (11) Agitation, (12) Loss of interest, (13) Indecisiveness, (14) Worthlessness, (15) Loss of energy, (16) Change in sleeping patterns, (17) Irritability, (18) Change in appetite, (19) Concentration difficulty, (20) Tiredness/fatigue, and (21) Loss of libido. Each item can be answered with a score from 0 to 3, with higher values corresponding to higher symptom load. The BSL-23 is a 23-item, self-report rating that measures BPD symptom severity within the last 7 days on a scale from 0 to 4 (Bohus et al., 2009). The 23 items include (1) Concentration difficulty, (2) Helplessness, (3) Absent-mindedness, (4) Disgust, (5) Thought of self-harm, (6) Distrust towards others, (7) Absent right to live, (8) Loneliness, (9) Stressful inner tension, (10) Fearful imagery, (11) Self-hatred, (12) Desire to punish self, (13) Shame, (14) Mood swings, (15) Voices inside/outside head, (16) Devastation by criticism (17) Vulnerability, (18) Fascination with death, (19) Senselessness, (20) Fear of loss of control, (21) Self-disgust, (22) Distance from oneself, and (23) Worthlessness. After the participant had filled out the questionnaires, around 1–2 pm in the afternoon, the saliva sample was collected and the experimental procedure proceeded as described elsewhere (Graumann et al., 2023). Testosterone assessment Salivary testosterone was collected using SaliCap devices (1.5 ml polypropylene tubes; IBL, Hamburg, Germany) at baseline, e.g. before study intervention, and stored at -80° until further analyses. Participants were not allowed to drink, eat or chew gum until 30 min before saliva collection. Testosterone levels were determined using a competitive immunoassay (IBL/TECAN, Hamburg, Germany) following the manufacturer’s recommendations. The detection limit was 2.1 pg/ml, precision parameters (coefficient of variation; CV) for medium concentrations at 50 pg/ml averaged below 5% CV for intra- and 10% CV for inter-assay variance. Statistical analyses Testosterone values were log transformed to fit normal distribution. Associations between testosterone levels and symptom severity were tested with regression analyses using SPSS 26. Because adipose tissue might influence the production of testosterone and affect depressive symptoms (Stanikova et al., 2019; Zhao et al., 2009), body-mass-index (BMI) was added as a predictor of no interest. To control for multiple comparisons in the by-item analyses, Bonferroni-correction was applied. Logistic regressions served to assess the influence of testosterone on presence/absence of BPD DSM-V criteria. Results Demographic Participants’ average age was M = 28 years (SD = 7, range 18-46). Average BMI was 23 kg/m2 (SD = 3). Sixteen women reported the intake of hormonal contraception. In naturally cycling women, calendar-based assessment indicated that n = 26 were in their follicular and n = 50 in their luteal phase, while in four women a tractable cycle was absent. Neither hormonal contraception (t(92) = 1.14, p = .31) nor cycle phase (F(2,89) = 0.59, p = .56) affected average testosterone levels. Forty-two women were smokers. Fifty women reported intake of at least one psychotropic substance, while 48 patients were free of psychotropic medication. Average sum scores were M = 27 (SD = 12) for BDI-II and M = 45 (SD = 20) for BSL-23. Associations between testosterone levels and symptom ratings Testosterone scores are presented for 94 individuals, in four patients testosterone values were missing because of measurement errors (e.g. the collected amount of saliva was too small for detection). The average testosterone levels of BPD patients were M = 30pg/ml (SD = 28pg/ml), ranging from 1 to 149pg/ml. BSL-23: Higher testosterone levels were associated with higher overall borderline symptom burden indicated by the BSL-23 sum score. When analysed by item, higher testosterone levels were associated with increased feeling of senselessness. Further items associated with higher testosterone were and self-disgust, disgust, worthlessness, distance from oneself, helplessness, self-hatred, absence of right to live, desire to punish self, and loneliness, however, these items were not significant after Bonferroni-correction (See Table 1). Table 1.Regression analyses show the influence of baseline salivary testosterone levels on Borderline symptom checklist (BSL-23) scores. Individual item (sorted by β values) and BSL-23 total scores are reported. Regr.-coeff.βTpuncorr(corr.Bonf)BSL-23 Items Senselessness (19).29.323.26.002*Self-disgust (21).31.313.11.003 Disgust (4).25.292.86.005 Worthlessness (23).27.282.77.007 Distance from oneself (22).21.252.48.015 Helplessness (2).19.242.38.019 Self-hatred (11).21.242.34.021 Absent right to live (7).21.222.19.031 Desire to punish self (12).22.222.15.035 Loneliness (8).17.202.04.045 Shame (13).17.191.91.059 Mood swings (14).16.181.72.088 Thought of self-harm (5).15.161.56.112 Vulnerability (17).12.161.54.127 Fascination with death (18).14.161.53.130 Concentration difficulty (1).12.151.46.147 Absent-mindedness (3).16.141.34.184 Fearful imagery (10).13.141.38.170 Fear of loss of control (20).12.131.31.193 Stressful inner tension (9).07.100.92.358 Devastation by criticism (16).07.070.72.477 Voices inside/outside head (15).02.030.29.775 Distrust towards others (6).01.010.12.905 BSL-23 total score.16.282.82.003** Note: Results of regression analyses, significance is reported as uncorrected two-tailed p-value and additionally marked as significant with applied Bonferroni correction to control for multiple comparisons. Questionnaire item numbers are reported in brackets. Table 1.Regression analyses show the influence of baseline salivary testosterone levels on Borderline symptom checklist (BSL-23) scores. Individual item (sorted by β values) and BSL-23 total scores are reported. Regr.-coeff.βTpuncorr(corr.Bonf)BSL-23 Items Senselessness (19).29.323.26.002*Self-disgust (21).31.313.11.003 Disgust (4).25.292.86.005 Worthlessness (23).27.282.77.007 Distance from oneself (22).21.252.48.015 Helplessness (2).19.242.38.019 Self-hatred (11).21.242.34.021 Absent right to live (7).21.222.19.031 Desire to punish self (12).22.222.15.035 Loneliness (8).17.202.04.045 Shame (13).17.191.91.059 Mood swings (14).16.181.72.088 Thought of self-harm (5).15.161.56.112 Vulnerability (17).12.161.54.127 Fascination with death (18).14.161.53.130 Concentration difficulty (1).12.151.46.147 Absent-mindedness (3).16.141.34.184 Fearful imagery (10).13.141.38.170 Fear of loss of control (20).12.131.31.193 Stressful inner tension (9).07.100.92.358 Devastation by criticism (16).07.070.72.477 Voices inside/outside head (15).02.030.29.775 Distrust towards others (6).01.010.12.905 BSL-23 total score.16.282.82.003** Note: Results of regression analyses, significance is reported as uncorrected two-tailed p-value and additionally marked as significant with applied Bonferroni correction to control for multiple comparisons. Questionnaire item numbers are reported in brackets. Regression analyses show the influence of baseline salivary testosterone levels on Borderline symptom checklist (BSL-23) scores. Individual item (sorted by β values) and BSL-23 total scores are reported. Note: Results of regression analyses, significance is reported as uncorrected two-tailed p-value and additionally marked as significant with applied Bonferroni correction to control for multiple comparisons. Questionnaire item numbers are reported in brackets. BDI-II: Higher testosterone levels were associated with overall higher BDI-II sum scores. By-item analysis revealed an association with higher sense of past failure, self-dislike and pessimism. Further items that showed a positive association but did not remain significant after Bonferroni-correction were suicidal thoughts or wishes, loss of pleasure, self-criticalness, worthlessness, guilt feelings, and indecisiveness (See Table 2). Table 2.Regression analyses show the influence of baseline salivary testosterone levels on Beck’s Depression Inventory II (BDI-II) scores. Individual item (sorted by β values) and BDI-II total scores are reported. Regr.-coeff.βTpuncorr(corr.Bonf)BDI Items Past failure (3).23.373.78.001*Self-dislike (7).23.343.45.001*Pessimism (2).22.333.26.002*Suicidal thoughts or wishes (9).12.272.67.009 Loss of pleasure (4).16.262.52.014 Self-criticalness (8).16.252.48.015 Worthlessness (14).16.252.47.015 Guilt feelings (5).16.242.32.022 Indecisiveness (13).17.222.16.033 Tiredness/fatigue (20).11.201.97.052 Punishment feelings (6).13.191.81.073 Agitation (11).12.191.84.068 Change in appetite (18).13.181.74.085 Crying (10).12.171.59.115 Loss of interest (12).10.161.53.110 Sadness (1).06.151.43.156 Change in sleeping patterns (16).10.151.39.167 Loss of energy (15).06.121.14.256 Concentration difficulty (19).07.121.20.235 Loss of libido (21).07.090.85.398 Irritability (17).04.070.60.549 BDI total score26.33.343.40.001*** Note: Results of regression analyses, significance is reported as uncorrected two-tailed p-value and additionally marked as significant with applied Bonferroni correction to control for multiple comparisons. Questionnaire item numbers are reported in brackets. Table 2.Regression analyses show the influence of baseline salivary testosterone levels on Beck’s Depression Inventory II (BDI-II) scores. Individual item (sorted by β values) and BDI-II total scores are reported. Regr.-coeff.βTpuncorr(corr.Bonf)BDI Items Past failure (3).23.373.78.001*Self-dislike (7).23.343.45.001*Pessimism (2).22.333.26.002*Suicidal thoughts or wishes (9).12.272.67.009 Loss of pleasure (4).16.262.52.014 Self-criticalness (8).16.252.48.015 Worthlessness (14).16.252.47.015 Guilt feelings (5).16.242.32.022 Indecisiveness (13).17.222.16.033 Tiredness/fatigue (20).11.201.97.052 Punishment feelings (6).13.191.81.073 Agitation (11).12.191.84.068 Change in appetite (18).13.181.74.085 Crying (10).12.171.59.115 Loss of interest (12).10.161.53.110 Sadness (1).06.151.43.156 Change in sleeping patterns (16).10.151.39.167 Loss of energy (15).06.121.14.256 Concentration difficulty (19).07.121.20.235 Loss of libido (21).07.090.85.398 Irritability (17).04.070.60.549 BDI total score26.33.343.40.001*** Note: Results of regression analyses, significance is reported as uncorrected two-tailed p-value and additionally marked as significant with applied Bonferroni correction to control for multiple comparisons. Questionnaire item numbers are reported in brackets. Regression analyses show the influence of baseline salivary testosterone levels on Beck’s Depression Inventory II (BDI-II) scores. Individual item (sorted by β values) and BDI-II total scores are reported. Note: Results of regression analyses, significance is reported as uncorrected two-tailed p-value and additionally marked as significant with applied Bonferroni correction to control for multiple comparisons. Questionnaire item numbers are reported in brackets. DSM-V BPD criteria: No effects were significant after Bonferroni correction. On a trend level, higher testosterone was associated with a higher rate of meeting the DSM-V criterion of self-destructiveness. Interestingly, the association between testosterone levels and uncontrollable anger was negative: patients who met this this criterion showed a tendency towards lower testosterone levels (See Table 3). Table 3.Logistic regression analyses of influence of basal testosterone levels on DSM-IV BPD criteria.DSM-V BPD criteriaRegr.-coeff. BWaldpuncorr (corr.Bonf)Fear of abandonment.2472.22.136 Unstable relations.1610.71.401 Identity disturbance.1210.75.386 Impulsivity-.0710.24.621 Self-destructiveness.7206.20.013 Affective lability-.7232.21.137 Chronic emptiness-.1250.39.533 Uncontrolled anger-.3153.87.049 Paranoia/dissociation-.1090.52.470 Table 3.Logistic regression analyses of influence of basal testosterone levels on DSM-IV BPD criteria.DSM-V BPD criteriaRegr.-coeff. BWaldpuncorr (corr.Bonf)Fear of abandonment.2472.22.136 Unstable relations.1610.71.401 Identity disturbance.1210.75.386 Impulsivity-.0710.24.621 Self-destructiveness.7206.20.013 Affective lability-.7232.21.137 Chronic emptiness-.1250.39.533 Uncontrolled anger-.3153.87.049 Paranoia/dissociation-.1090.52.470 Logistic regression analyses of influence of basal testosterone levels on DSM-IV BPD criteria. Discussion Our findings show a positive association between salivary testosterone levels and borderline and depressive symptoms in women with BPD. In particular, patients with higher testosterone levels showed higher ratings of self-dislike, senselessness and pessimism, and the feeling of being a failure. Results from both BSL-23 and BDI-II were highly consistent, which speak for a well-defined cluster of internalising symptoms that correlates with testosterone levels. In the clinician administered interview we found a non-significant trend towards stronger self-oriented destructiveness (self-harming and suicidal behaviour), whereas other-oriented anger and aggression showed a trend towards a negative association with baseline testosterone levels. Our results do not support the hypothesis that testosterone in female BPD patients promotes externalising behaviour. Instead, the strongest associations were found for internalising symptoms characterized by a pessimistic and derogatory view of oneself. This symptom cluster overlaps between BPD and MDD symptomatology. In contrast, autonomous (sleep, appetite), motivational (apathy, agitation) or cognitive (concentration) symptoms of MDD were not significantly associated with testosterone levels. Taken together, higher salivary testosterone in female BPD patients is associated with a negative and pessimistic affect and cognition towards the self. This combination of negative affect and dysfunctional beliefs of worthlessness, to which BPD patients are prone, might further promote self-harm and suicidality. Consistently, higher testosterone levels have been associated with a higher number of previous suicide attempts in patients with bipolar disorder (Sher, 2012). Androgen receptors, to which testosterone binds, show a high expression in the amygdala (Simerly et al., 1990), and testosterone administration increases amygdala reactivity in young and middle-aged women (Bos et al., 2013; Van Wingen et al., 2009). This might suggest a potential mechanism of action, as amygdala hyperactivation is a stable finding in BPD patients (Donegan et al., 2003). More research is needed to understand the neuroendocrinological trajectories of BPD development that lead to increased testosterone levels in adulthood. Our findings suggest a link between testosterone and depressive and borderline symptomatology that is not primarily driven by patient BMI. Furthermore, in our sample oral contraception intake or cycle phase did not affect average testosterone levels. However, testosterone is known to fluctuate within the menstrual cycle, with an increase during ovulation (Al-Dujaili & Sharp, 2012). This calls for further investigations of dynamic changes of BPD symptoms across the menstrual cycle and the interplay with other gonadal hormones (Eisenlohr-Moul et al., 2018). Suicidal behaviour, which is common in BPD, changes throughout the menstrual cycle (Saunders & Hawton, 2006), and successful treatment and prevention requires a better understanding of potentially involved neuroendocrinological factors (Owens & Eisenlohr-Moul, 2018). Identifying different phenotypes of androgen receptor sensitivity and circulating testosterone levels might also help to account for discrepant findings between MDD and BPD patients. In MDD, testosterone levels are reported to be reduced, and lower levels predict higher disease burden. While testosterone administration shows anti-depressive effects (Zito et al., 2023), age and, importantly, sex seem to be relevant mediators. In female samples, RCTs of testosterone administration did not significantly decrease depressive symptoms in MDD (Dichtel et al., 2020) or anorexia nervosa (Kimball et al., 2019). Delineating the specific symptoms that respond to external testosterone administration is relevant for these patient groups. In general, more research is needed to better understand potentially discrepant findings of MDD and BPD populations and the effect of biological sex. Furthermore, it would be interesting to know which specific depressive symptoms drove previously reported positive or negative associations with testosterone in MDD, and whether other clinical conditions such as PTSD and anxiety disorders show similar effects. So far, our findings do not allow the conclusion that the observed associations between testosterone and internalising symptoms are specific to women with BPD. Another factor to consider in future research is stress, which constitutes both a cause for and a consequence of clinical symptom load. Testosterone increases after acute stress in patients with BPD and healthy controls (Deuter et al., 2021), as testosterone production is regulated by the hypothalamic-pituitary-gonadal which is bidirectionally connected with the hypothalamic–pituitary–adrenal axis (Acevedo-Rodriguez et al., 2018; Viau, 2002). Do BPD patients show higher testosterone levels due to higher daily stress that stems from symptom load? Or do higher testosterone levels precede borderline symptomatology and constitute a potential risk factor? Does borderline symptomatology remit when testosterone levels decrease, potentially pointing towards treatment approaches? Only longitudinal or interventional studies can answer these questions by targeting the causal direction of the association between testosterone and symptom load demonstrated by the present cross-sectional analysis. In contrast to increasing testosterone levels in MDD, our findings suggest that women with BPD might benefit from anti-androgen treatment which reduces testosterone levels. RCTs are lacking, however, a single case study reported that hyperandrogenism/PCOS treatment led to the remission of BPD symptoms in a young woman (Trisno et al., 2016). Reversely, our findings might invite more careful monitoring of psychiatric side effects of androgen treatment, especially in patients with known BPD symptoms. Limitations The present study only included women (female sex assigned at birth) with BPD, therefore, the association between testosterone on BPD symptoms in males remains unknown. Testosterone analyses were based on one sample. Some patients were smokers or taking psychiatric medication, which might influence endocrinological status. Conclusion Our findings suggest that salivary testosterone levels are associated with a relatively narrow cluster of internalising symptoms that overlap between BPD and MDD, including feelings of self-dislike, senselessness and pessimism, and the feeling of being a failure. More research is needed to better understand the influence of gonadal hormones on psychiatric symptomatology. In women, the influence of cycling hormones has been considered a nuisance factor and therefore symptomatically neglected for a long time. Elucidating the influence of (cycling) gonadal hormones can promote the development of cycle-sensitive female mental health approaches to improve prevention and treatment.