Psychiatry Research Communications 3 (2023) 100117

Contents lists available at ScienceDirect

Psychiatry Research Communications

journal homepage: www.sciencedirect.com/journal/Psychiatry-Research-Communications

Hypomodulation of salivary oxytocin in patients with borderline personality

disorder: A naturalistic and experimental pilot study
T. Aboulafia-Brakha a, *, N. Perroud a, D. Suchecki b, R. Nicastro a, K. Dieben a, L. Curtis a
a
Division of Psychiatric Specialties, Department of Psychiatry, Geneva University Hospitals, Geneva, Switzerland
b
Department of Psychobiology, Universidade Federal de S~ ao Paulo, S~ao Paulo, Brazil

A R T I C L E I N F O A B S T R A C T

Handling Editor: Dr. Leonardo Fontenelle We report here a pilot trial that mainly aimed to assess the endogenous secretion of oxytocin (OXT) in patients
with Borderline personality disorder (BPD) and healthy controls. It is the first trial with BPD using salivary OXT
Keywords: and studying its reactivity in a natural setting and an experimental stress task. Compared to controls, patients with
Borderline personality disorder BPD showed lower variation of OXT in the natural setting and lower OXT reactivity during the stress task,
Oxytocin
contrasting to higher perceived stress and anger states. We confirm the feasibility of the protocol. Our results
Cortisol
encourage the implementation of a larger trial to address specific hypotheses.
Stress

1. Introduction assessing OXT reactivity to stress in BPD patients could provide a better
understanding of neurobiological mechanisms implicated in the disease's
Interest in the therapeutic effects of intranasal oxytocin (OXT) on symptomatology.
social appraisal of patients with borderline personality disorder (BPD) We report here a pilot trial in which we could capture and compare
has grown in the last years, but knowledge about endogenous OXT in this OXT variations in everyday life and during a stress task in patients with
population is contrastingly poor (Bertsch and Herpertz, 2018). The few BPD and controls. Assessment at different time-points was possible
existing studies regarding basal levels report mixed findings, with either thanks to recent reliable quantification of OXT in saliva samples. In
lower plasma or serum levels of OXT and reduced expression of OXT addition, we measured salivary cortisol under the same conditions to
receptor in BPD patients compared to controls (Bertsch et al., 2013; Ebert ensure reliability of samples (due to the well documented cortisol
et al., 2018; Carrasco et al., 2020) or no differences between groups secretion pattern in naturalistic and experimental settings), but also
(Bomann et al., 2017; Bonfig et al., 2022). However, differences in because one of the goals of a future larger trial is to additionally assess the
plasma OXT levels may be more evident when analysing subgroups, such interplay between these two hormones in BPD patients, as reported for
as BPD patients with disorganised attachment (Jobst et al., 2016), or healthy populations (Jong et al., 2015; Bernhard et al., 2018; Alley et al.,
when OXT is studied in a more dynamic way, for example, in response to 2019).
social interaction tasks. In fact, patients with BPD may show decreased
plasma OXT in response to both social exclusion and proximity (Jobst 2. Methods
et al., 2016; Bonfig et al., 2022).This apparent paradox could be poten-
tially related to the stressful nature of social interaction per se, but OXT 2.1. Sample characteristics
reactivity to stress in patients with BPD has never been studied.
BPD is a highly prevalent and debilitating psychiatric condition Nineteen female participants aged between 18 and 25 years-old
mainly characterised by intense emotion dysregulation and interpersonal completed the entire study protocol; of these, 10 patients with BPD
dysfunction, with fear of abandonment and attachment problems (As- were recruited in two outpatients units of the Service of Specialised
sociation, 2013). Symptoms are mainly triggered or emphasized during Psychiatry of the Geneva University Hospitals and nine controls with no
stressful situations or perceived stress. As OXT is not only an attachment current or previous history of psychiatric disorder were recruited by
hormone, but also tends to increase in healthy subjects during stressful advertisement. Patients met criteria for BPD as defined by the DSM-5 and
situations (Jong et al., 2015; Bernhard et al., 2018; Alley et al., 2019), assessed through the structured clinical interview for DSM-5 personality

* Corresponding author. Unit of Psychiatry, Geneva University Hospitals, Switzerland.

E-mail address: Tatiana.aboulafi[email protected] (T. Aboulafia-Brakha).

https://doi.org/10.1016/j.psycom.2023.100117
Received 17 November 2022; Received in revised form 7 February 2023; Accepted 8 March 2023
2772-5987/© 2023 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
T. Aboulafia-Brakha et al. Psychiatry Research Communications 3 (2023) 100117

disorders (First et al., 2015). Control participants were also screened for this moment, participants also returned the evening saliva samples
BPD during an interview. They were excluded if they met more than collected previously at home.
three DSM-5 BPD criteria, five being the threshold for diagnosis.
Exclusion criteria for all participants: active comorbidities such as 2.2.1. Assessment in the participant's naturalistic environment
moderate or severe major depressive disorder, formally established Salivary cortisol levels (ng/mol) were measured at six different time
post-traumatic stress disorder, psychotic disorders, substance use/abuse points throughout the day on a week-day: three time-points in the
disorders, brain injury, neuroendocrine diseases or hormonal replace- morning, with a time sampling schedule starting at awakening, then 30
ment. No participants had undergone major stressful life events (such as and 45 min later. The fourth measure was collected between 12 and 2
loss of a close relative, loss of employment, change in marital status or in p.m. in our laboratory, and the two remaining samples at 6 p.m. and
medical condition) in the two months prior to enrolment in the study. before going to sleep. Participants received a reminder the evening prior
As shown in Table 1, the groups did not differ significantly regarding to data collection in order to maximize adherence to the protocol, and
age or attachment patterns. However, they differed regarding self- were asked to register the precise time of collection. Two participants
reported adverse childhood experiences and clinical symptomatology who were not able to follow the morning schedule (one BPD patient and
such as depressive symptoms, trait anxiety, trait-anger, post-traumatic one control) were asked to restart circadian assessment on a different
stress self-reported symptoms, with, as expected, higher scores in BPD day.
patients. Of note, four patients were treated with SSRI and four partici- Salivary OXT (pg/ml) was measured twice, with one sample at
pants were taking hormonal contraceptives. The remaining participants awakening (together with the first cortisol awakening sample when
were assessed in their presumed luteal phase, at least 14 days following participants were still in bed) and the second at mid-day in our labora-
the beginning of their menstrual cycle. tory. As OXT samples need to be stored at – 20oC within a few hours in
order to maintain OXT stability, we could not collect an evening sample.
Concomitantly with the mid-day hormonal measures participants
2.2. Study procedures
assessed their momentary perceived psychological state with stan-
dardized questionnaires for stress and anxiety (STAI-S (Marteau and
This study was registered on clinicaltrials.gov (NCT05357521) and
Bekker, 1992))and for anger feelings (STAXI-S (Spielberger et al., 1995)).
carried out in accordance with the code of Ethics of the World Medical
Association (declaration of Helsinki involving Humans). Approval by the
2.2.2. Assessment during a stress task
local ethical committee was obtained (Swissethics NAC2002-00067) and
The Trier Social Stress Test (TSST) (Labuschagne et al., 2019) was
a financial compensation was given to participants after completion of
administered according to standard procedures except for the panel
the entire protocol. We collected data (salivary cortisol and OXT) in the
which was composed of only one person. Experimental sessions were
participants’ naturalistic environments (section 2.2.1) and during an
scheduled to start between 1.30 and 2 p.m. to avoid circadian variability.
experimental session (section 2.2.2) that took place on a different day.
Participants had their meal no later than 12.30 p.m. Upon arrival at the
Each participant completed three visits to our laboratory, all scheduled
laboratory they were installed in a quiet room, drank a fruit juice con-
between 12 and 2 p.m. in order to control for circadian hormone varia-
taining 11 g of sugar, and rested during 20 min with no possibility of
tions. During the first visit participants were screened for inclusion and
using their mobile phones or any other form of communication. Baseline
exclusion criteria, were informed of the procedures and gave informed
samples were then collected for cortisol, OXT and self-reported stress and
consent. They also received self-report questionnaires and material for
anxiety state (STAI-S) and anger state (STAXI-S). The stress part took
collecting saliva, which they returned at the second visit. The latter was
place in a different room following a 5 min preparation period and a
held the day participants collected saliva in their naturalistic environ-
saliva sample for measuring cortisol. The task consisted of a 5 min speech
ment (morning and evening samples were collected at home but the mid-
for a fictitious job interview and a 5 min surprise arithmetic task,
day sample was collected in our laboratory with a brief self-report state
following which post-task measures were taken for cortisol, OXT and
assessment; this also ensured compliance with the protocol and allowed
self-reported states. Specific measures were taken during recovery at 15
adequate storage of samples). Finally, the third visit was scheduled for
min (cortisol and OXT), 30 min (only cortisol) and 50 min post-task
the experimental procedure one to three days after the second visit. At
(cortisol, OXT and self-reported measures) while participants remained
quietly in the resting room.

Table 1 2.2.3. Quantification of salivary OXT and cortisol

Demographic and clinical self-report questionnaires data. Saliva was collected using Salivettes with synthetic swab (Sarstedt,
Mean (SD) t (17) P value Germany) chewed for at least 1min. Swabs were stored at
20oC until
BPD Controls analysis and temperature was maintained during shipping. The same
tube allowed for both OXT and cortisol quantifications although analyses
Age (years) 20.9 (2.13) 21.78 (2.9)
0.75 0.45
Adverse childhood experiences 5.20 (2.78) 1.44 (1.88) 3.41 0.003
were performed by different laboratories. OXT concentrations were
(ACE) quantified by radioimmunoassay (RIAgnosis, Sinzing, Germany)
Depressive symptoms (BDI) 30.60 6.4 (4.2) 5.82 <0.001 following the procedures described in previous studies (Jong et al., 2015;
(11.76) Bernhard et al., 2018). Cortisol was quantified by immunoassay with a
Traumatic stress symptoms 49.8 11.89 6.37 <0.001
Cobas e 411 analyser from Roche, with an interassay variability of
(PCL) (14.94) (10.25)
Anxiety-trait (STAI-T) 63.7 (7.79) 39.89 6.33 <0.001 <6.6%.
(8.86)
Anger-trait (STAXI-T) 25.3 (6.31) 15.4 (2.45) 4.39 <0.001 2.3. Main statistical procedures
Attachment styles (RSQ)
Secure 2.78 (0.51) 3.2 (0.48)
1.85 0.08
Dismissing 3.42 (0.76) 3.64 (0.69)
0.68 0.51 Statistical analyses were carried out using the SPSS 28.0 statistics
Preoccupied 3.27 (0.48) 2.61 (0.96) 1.94 0.07 software. Self-report measures during experimental tasks (Fig. 1c and d)
Fearful 3.20 (1.11) 2.97 (0.52) 0.56 0.59 were analysed with a mixed 2  3 repeated measure analysis of variance
ACE Adverse Childhood Experiences; BDI Beck Depression Inventory; PCL Post- (rAnova), with group (BPD, controls) as between factor, and condition
TraumaticStress Disorder Check-List, RSQ Relationship Questionnaire, STAI-T (baseline, post-stress and recovery) as within factor. Naturalistic and
State-Trait Anxiety Inventory – Trait Subscale; STAXI-T State-Trait Anger In- experimental OXT measures were normally distributed (Shapiro Wilk test
ventory – Trait Subscale. >0.05). However, naturalistic and experimental cortisol measures failed

2
T. Aboulafia-Brakha et al. Psychiatry Research Communications 3 (2023) 100117

Fig. 1. A and B Cortisol and oxytocin levels of patients with borderline personality disorder (BPD) and controls in their natural environment. Fig. 1 C,D,E and F. Self-
reported stress and anger as well as cortisol and oxytocin levels during the stress task for BPD patients and controls.

to meet assumption of normality (p < 0.05), with positively skewed data. p < 0.01) and self-reported anger state (STAXI-S) (BPD: M ¼ 20.6, SD ¼
Therefore, cortisol values were log-transformed for analyses (with a 6.70, controls: M ¼ 15.5, SD ¼ 1.33, t (17) ¼ 2.12, p ¼ 0.04). The
constant due to values inferior to 1). Log-transformed cortisol values MANCOVA yielded a significant between-group difference for both
were normally distributed. For cortisol and OXT naturalistic (Fig. 1a and hormones combined, with a main effect of group irrespective of hor-
b) and experimental data (Fig. 1e and f) we computed the area under the mone, F (2, 15) ¼ 5.47, p ¼ 0.01, partial η2 ¼ 0.42, power ¼ 77%.
curve with respect to increase (AUCi) (in order to capture change over Further, the main effect of group was significant for both OXT (F (1, 16)
time) (Pruessner et al., 2003). We then performed multivariate analyses ¼ 7.91, p ¼ 0.01, partial η2 ¼ 0.33) and cortisol F (1, 16) ¼ 4.38, p ¼
of covariance (MANCOVA) for naturalistic data with Cortisol AUCi and 0.05, partial η2 ¼ 0.21), with BPD patients showing lower variation
OXT AUCi as dependent variables, while controlling for childhood compared to controls, indicating lower cortisol decrease throughout the
adverse experiences with ACE scores a covariate, due to its known effects day and lower OXT increase in the morning.
on both basal cortisol and OXT. For experimental data, we performed a
Multivariate analysis of Variance (MANOVA) with cortisol AUCi and OXT
AUCi as dependent variables. For both analyses Box's Test of Equality of 3.2. Experimental setting results
Covariance showed that assumption of homogeneity of covariance
was met (p > 0.05); assumption of homogeneity of variance was also The MANOVA showed no main effect of group irrespective of hor-
met for each outcome variable, as shown by Levene's Test of mone, F (2, 15) ¼ 2.41, p ¼ 0.12, partial η2 ¼ 0.21; however, while the
Equality of Error Variances (p > 0.05). effect of group was significant for OXT (F(1, 17) ¼ 4.25, p ¼ 0.05, partial
η2 ¼ 0.20), with greater OXT increase in controls compared to BPD pa-
3. Results tients, it did not reach significance for cortisol F (1, 17) ¼ 0.46, p ¼ 0.50,
partial η2 ¼ 0.02),. For the self-report data, the 2 (group) X 3 (condition)
3.1. Naturalistic environment repeated measure ANOVA on the STAI-S revealed a main effect of con-
dition (F (2, 34) ¼ 54.8, p < 0.001 partial η2 ¼ 0.76) and a main effect of
Groups did not differ regarding awakening time (BPD: M ¼ 8.6, SD ¼ group (F (1, 17) ¼ 17.73, p < 0.001, partial η2 ¼ 0.51), but no significant
1.11; controls: M ¼ 7.94, SD ¼ 1.32, t (17) ¼ 1.5, p ¼ 0.35). Compared to interaction; as for STAXI-S the same pattern was observed with a main
controls, BPD patients showed higher self-perceived stress-state (STAI-S) effect of group (F (1, 17) ¼ 4.58, p ¼ 0.05, partial η2 ¼ 0.21) and con-
(BPD: M ¼ 50.2 SD ¼ 11.47, controls: M ¼ 30.5 SD ¼ 5.41, t (17) ¼ 4.69, dition (F (2, 34) ¼ 8.96 p < 0.001, partial η2 ¼ 0.34).

3
T. Aboulafia-Brakha et al. Psychiatry Research Communications 3 (2023) 100117

3.3. Correlations Credit roles

No significant correlations were detected between OXT AUCi and Log TAB: conceptualization, funding acquisition, project administration,
cortisol AUCi during natural or experimental settings. Significant nega- methodology, data acquisition, formal analysis and writing; NP and LC:
tive correlations were found between OXT AUCi in the natural setting methodology, resources and editing; DS: conceptualization, methodol-
and scores in BDI (r ¼ - 0.72, p < 0.001), PCL (r ¼ - 0.75, p < 0.001), ACE ogy, and editing; RN and KD clinical assessment, investigation.
(r ¼ - 0.51, p ¼ 0.02), STAI-T (r ¼ - 0.62, p < 0.01) and STAXI-T (r ¼ -
0.58, p < 0.01); OXT AUCi in experimental setting did not show signif-
Declaration of competing interest
icant positive correlation with psychological variables, but with the
naturalistic OXT AUCi (r ¼ 0.67, p < 0.01).
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influence
4. Discussion
the work reported in this paper.

Repeated measurements under different contexts showed lower OXT

Acknowledgement
modulation in BPD patients compared to controls, with lower variation in
naturalistic settings and lower reactivity during a stress task, contrasting
We thank Radek Ptak for his comments on the manuscript; Alexis Giff
the higher perceived stress and anger states. The OXT pattern of stress
for assisting with some aspects of data collection and Noemi Koenzi for
reactivity observed in controls is consistent with previous findings in
her help with databasis.
healthy participants (Jong et al., 2015; Bernhard et al., 2018; Alley et al.,
2019). The altered reactivity of OXT in BPD patients is also consistent
References
with previously reported dysfunction of plasma OXT in this population
(Jobst et al., 2014; Bonfig et al., 2022). However, our study brings Alley, J., Diamond, L.M., Lipschitz, D.L., Grewen, K., 2019. Associations between oxytocin
additional information on stress reactivity, leading us to postulate that and cortisol reactivity and recovery in response to psychological stress and sexual
the inability to upregulate OXT during stressful conditions could underlie arousal. Psychoneuroendocrinology 106, 47–56.
Association, A.P., 2013. Diagnostic and Statistical Manual of Mental Disorders, fifth ed.
a maladaptive stress response (higher perceived stress for example) and
(Arlington).
emotion dysregulation. Strong negative correlations found between Bernhard, A., van der Merwe, C., Ackermann, K., Martinelli, A., Neumann, I.D.,
naturalistic OXT and psychological self-reported stress, anxiety and anger Freitag, C.M., 2018. Adolescent oxytocin response to stress and its behavioral and
endocrine correlates. Horm. Behav. 105, 157–165.
traits gives additional support to this hypothesis. Further, our findings
Bertsch, K., Herpertz, S.C., 2018. Oxytocin and borderline personality disorder. Curr Top
could bring useful information for designing trials aiming to assess the Behav Neurosci 35, 499–514.
effects of intranasal OXT on stress reactivity in patients with BPD. Bertsch, K., Schmidinger, I., Neumann, I.D., Herpertz, S.C., 2013. Reduced plasma
Regarding cortisol levels, the BPD group showed lower circadian oxytocin levels in female patients with borderline personality disorder. Horm. Behav.
63 (3), 424–429.
release of cortisol, but we failed to detect significant differences in Bomann, A.C., Jorgensen, M.B., Bo, S., Nielsen, M., Gede, L.B., Elfving, B., Simonsen, E.,
cortisol reactivity. This could either be due to the lack of statistical power 2017. The neurobiology of social deficits in female patients with borderline
or to the diminished stress intensity of the task with a reduced panel of personality disorder: the importance of oxytocin. Pers. Ment. Health 11 (2), 91–100.
Bonfig, J., Herpertz, S.C., Schneider, I., 2022. Altered hormonal patterns in borderline
experts. This should be reviewed for a larger trial, especially if we aim to personality disorder mother-child interactions. Psychoneuroendocrinology 143,
assess the hormonal interplay in BPD patients (Labuschagne et al., 2019). 105822.
Indeed, in healthy subjects, cortisol seems to be modulated by OXT, with Carrasco, J.L., Buenache, E., MacDowell, K.S., De la Vega, I., Lopez-Villatoro, J.M.,
Moreno, B., Diaz-Marsa, M., Leza, J.C., 2020. Decreased oxytocin plasma levels and
a particular pattern of interaction between these hormones. Experiments oxytocin receptor expression in borderline personality disorder. Acta Psychiatr.
showed that OXT peaks before cortisol and that up-regulation of OXT is Scand. 142 (4), 319–325.
related to down-regulation of cortisol in the recovery period (Engert Ebert, A., Edel, M.A., Gilbert, P., Brune, M., 2018. Endogenous oxytocin is associated with
the experience of compassion and recalled upbringing in Borderline Personality
et al., 2016; Bernhard et al., 2018; Alley et al., 2019). Finally, a larger Disorder. Depress. Anxiety 35 (1), 50–57.
trial assessing cortisol and OXT reactivity to stress in patients with BPD Engert, V., Koester, A.M., Riepenhausen, A., Singer, T., 2016. Boosting recovery rather
would also allow analyses of specific covariates, even though we could than buffering reactivity: higher stress-induced oxytocin secretion is associated with
increased cortisol reactivity and faster vagal recovery after acute psychosocial stress.
already control for childhood adverse experiences, which considerably
Psychoneuroendocrinology 74, 111–120.
influences OXT and cortisol release during adulthood (Bertsch et al., First, M.W., JBW, Benjamin, L.S., Spitzer, R.L., 2015. User's Guide for the SCID-5-PD
2013). (Structured Clinical Interview for DSM-5 Personality Disorder). Arlington, VA.
To conclude, this is the first study assessing salivary OXT in BPD Jobst, A., Albert, A., Bauriedl-Schmidt, C., Mauer, M.C., Renneberg, B., Buchheim, A.,
Sabass, L., Falkai, P., Zill, P., Padberg, F., 2014. Social exclusion leads to divergent
patients. The different pattern of OXT reactivity observed in these pa- changes of oxytocin levels in borderline patients and healthy subjects. Psychother.
tients brings us a step forward in the elucidation of the neurobiological Psychosom. 83 (4), 252–254.
underpinnings of BPD. Hence, this pilot study clearly encourages the Jobst, A., Padberg, F., Mauer, M.C., Daltrozzo, T., Bauriedl-Schmidt, C., Sabass, L.,
Sarubin, N., Falkai, P., Renneberg, B., Zill, P., Gander, M., Buchheim, A., 2016. Lower
implementation of a larger trial. oxytocin plasma levels in borderline patients with Unresolved attachment
representations. Front. Hum. Neurosci. 10, 125.
Ethical considerations Jong, T.R., Menon, R., Bludau, A., Grund, T., Biermeier, V., Klampfl, S.M., Jurek, B.,
Bosch, O.J., Hellhammer, J., Neumann, I.D., 2015. Salivary oxytocin concentrations
in response to running, sexual self-stimulation, breastfeeding and the TSST: the
This study was carried out in accordance with the code of Ethics of the Regensburg Oxytocin Challenge (ROC) study. Psychoneuroendocrinology 62,
World Medical Association (declaration of Helsinki involving Humans). 381–388.
Labuschagne, I., Grace, C., Rendell, P., Terrett, G., Heinrichs, M., 2019. An introductory
Approval by the local ethical committee was obtained (Swissethics guide to conducting the trier social stress test. Neurosci. Biobehav. Rev. 107,
NAC2002-00067). Informed consent was obtained from all participants. 686–695.
The study was also registered on clinicaltrials.gov (NCT05357521). Marteau, T.M., Bekker, H., 1992. The development of a six-item short-form of the state
scale of the Spielberger State-Trait Anxiety Inventory (STAI). Br. J. Clin. Psychol. 31
(3), 301–306.
Funding source Pruessner, J.C., Kirschbaum, C., Meinlschmid, G., Hellhammer, D.H., 2003. Two formulas
for computation of the area under the curve represent measures of total hormone
Geneva University Hospitals- Research and Development grant (PRD- concentration versus time-dependent change. Psychoneuroendocrinology 28 (7),
916–931.
HUG 05-2022-II) accorded to TAB. The funding source had no role in the Spielberger, C.D., Reheiser, E.C., Sydeman, S.J., 1995. Measuring the experience,
conception or accomplishment of the study. expression, and control of anger. Issues Compr. Pediatr. Nurs. 18 (3), 207–232.