The present study is planned as a randomized controlled treatment trial (trial design ), involving a guideline-based psychiatric treatment (General Psychiatric Management; GPM; (20)) for BPD, over the course of a 4-month-plus-12-month-follow-up treatment program, in comparison with a 4-month treatment as usual, by focusing on the underlying mechanisms of change. The outcome part is a superiority trial, while the mechanisms part focuses on controlled comparison of processes of change. The trial takes place in the context of a University Hospital (study setting ).
Ethical Clearance
With respect to the submitted project on the SwissEthics plateform, the competent Canton de Vaud Ethics Committee, has approved the study (2017–02167). The study is registered (NCT0317818). In keeping with the established, and approved, Data Management Plan, only anonymous data will be kept in the file. All video raw data, where the patient may be identified, as well as fMRI data, will be stored separately from patient identifiers and from the main dataset.
Inclusion Criteria Of The Patients
Patients with Borderline Personality Disorder (BPD) according to Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (73) mastering French to a sufficient extent will be included. Rate of comorbidity is expected to be high. All patients accepting the study by informed written consent will be included in the outcome and mechanisms parts of the study; a sub-group of patients with additional inclusions criteria (female, younger than 45 years, right-handed, no or stable medication and absence of formal counter-indication on the security check) will be included in the fMRI part of the study.
Exclusion Criteria Of The Patients
Patients with neurocognitive disorder, psychosis and bipolar disorder I will be excluded from the trial. In order to ensure generalizability of the results to a wide variety of clinical settings, no other exclusion criteria will be applied.
Patients And Therapists
Based on a power-analysis (presumed power .819, for 2 concomitant mechanisms, d = 0.60; two-tailed alpha = .05; 30% drop-out), N = 80 patients presenting with BPD are planned to be recruited for the present study at the Department of Psychiatry, University of Lausanne, Switzerland. In total, N = 10 therapists (psychiatrists or psychologists) participate in the study for the GPM treatment; they have at least 6 hours of specific training in psychiatric treatment for BPD (74), in addition to training in psychotherapy according to Federal regulations.
Randomization Procedures
Block randomization (blocks of 10) will be used throughout the study. The randomization will be performed online by an independent researcher, who prepares sealed enveloped informing about the attribution to one out of two conditions. Informed consent will be obtained by a trained researcher (informed consent ). Upon inclusion into the protocol, the envelope will be opened and the condition will be communicated to patient and therapist. The (outcome and mechanisms) assessors and data analysts will be blind to the condition, but the patients nor the therapist will not be. In order to control for biases, assessors will be polled at the end of the study with regard to each patient’s condition . No a priori stratification will be performed.
Interventions
Brief treatment encompasses the communication about diagnoses, problem areas, anamnesis, the work on treatment focus, objectives and motivation, the treatment of treatment-interfering problems and the elaboration of interpretations related with the core concept of interpersonal hypersensitivity, according to the principles of General Psychiatric Management for BPD (GPM; (20, 75–77). Therapist adherence to the protocol will be self-assessed by therapist using Gunderson’s (2016) questionnaire for adherence to good psychiatric management principles, to be filled in after each treatment . For the TAU, 10 therapists will intervene using non-specific crisis management as usual (minimal ethical assessment and contact with the patient according the local directives, ensuring safety management). Should a participant request discontinuation of the intervention, the study therapist will use this as criteria for discontinuing and modifying the allocated intervention. Relevant concomitant care was permitted during the trial, and was recorded .
Assessments
Assessments take place at intake, 2 months and discharge, plus at follow-up after 12 months. At follow-up, the participants will be contacted by the research team .
Instruments And Materials
The main outcome measure is the Zanarini Rating Scale for Borderline Personality Disorder (ZAN-BPD (78)). ZAN-BPD is a continuous hetero-administered measure assessing the nine criteria outlined in DSM-5, on a continuous Likert-type scale ranging from 0 to 4. As such, it yields a total theoretical score of 36. A comprehensive validation study has shown its reliability, validity and sensitivity to change (78).
Emotion processing related to self-criticism will be assessed using the self-criticism task. This task involves two main steps. (1) Conduct of a two-chair dialogue on self-criticism, an individualized and therefore particularly emotion-arousing procedure ((35, 55, 79) and observer’s process rating of the patient’s emotions using the Classification of Affective Meaning States (CAMS;(80); see also (34, 47)) with the aim of extracting n = 20 individualized self-critical words for each patient at each assessment point. Increase in emotion flexibility (i.e., more different emotion categories as a reaction to the self-criticism) over time, along with a decrease in self-contemptuousness associated with the self-criticism over time, are indicators of productive change. (2) Test, one week later, of neural correlates of the processing related to the n = 20 individualized self-critical words (extracted from step 1), in comparison with a set of n = 20 negative emotional (81), n = 20 positive emotional (81), n = 20 neutral words and n = 20 non-words (symbols; in total N = 100 words; presented in a random order). Self-reported emotional arousal (on SAM) and self-esteem (on SSES), along with observer-reported fear/shame (on CAMS) reaction, will be measured as manipulation checks for step 1; SAM assessment will be done for each stimulus presented in the scanner. This task was empirically tested in a non-published pilot study: for n = 5 individuals, we showed for the individualized words higher subjective arousal levels than for the standardized negative words, along with comparable neurofunctional activations.
Change in socio-cognitive processing of interpersonal patterns will be assessed using two independent tasks, one behavioral, and the other fMRI. The behavioral task involves the conduct of a structured interview using the Relationship Anecdote Paradigm (67) and based on this video-taped structured material, observer’s process rating of the patient narrative using the CCRT (67–69) with the aim of comparing its pervasiveness pre-post-follow-up (see Fig. 1). Decrease in CCRT pervasiveness is an indicator of productive change. The fMRI task involves the appreciation of humoristic stimuli measuring the patient’s theory of mind; this task has previously been validated for BPD ((70); see for the validation of the stimuli (82, 83)). It involves the processing and understanding of three sets of stimuli, presented in a pseudo-random order: (1) ToM (theory of mind): visual jokes requiring attributing false mental states to the protagonists presented in the cartoons (30 stimuli), (2) PUN (visual puns): visual puns, i.e., cartoons that are based on visual similarities, not requiring attributing false mental states (30 stimuli) and (3) a non-humorous control condition with incongruent visual information (30 stimuli, in total N = 90). Manipulation checks involve the assessment of the understanding of each joke. Decrease in activation of ToM network over time is an indicator of productive change.
The remaining assessments concern secondary outcomes and include the Outcome Questionnaire-45 (84) which is a self-report questionnaire comprising 45 items aiming at assessing psychotherapeutic results, including a global score and three sub-scale scores: symptomatic level, interpersonal relationships and social role. It has been translated and validated in French (85). The Borderline Symptom List (BSL-23, (86)) is a self-report questionnaire assessing the BPD symptomatology using 23 items ; it represents a short version of the more extensive Borderline Symptom List (87) for which excellent psychometric properties were reported. Similar results were found for the short version used in this study (86).The French version has shown comparable validity coefficients (88). The Inventory of Interpersonal Problems (IIP, (89); French translation by Stigler) is a self-report questionnaire assessing interpersonal patterns on several dimensions, such as affirmation, affiliation, submission, intimacy, responsibility and control. In total, this questionnaire comprises 64 items. Spielberger State-Trait Anger Inventory (STAXI-2;(90)) is a self-report questionnaire on trait and state anger, using 44 items. The French validation and adaptation was carried out by Borteyrou, Bouchon-Schweitzer and Spielberger (91). The Difficulties in Emotion Regulation Scale (DERS; (92)) is a self-report questionnaire assessing the quality of emotion regulation using 36 items. The French translation and validation of this instrument yielded satisfactory factor structure on a student sample (93). As measure of the therapeutic alliance, the Working Alliance Inventory (WAI; (94); French validation (95)) will be given after each session; the therapeutic alliance will be introduced as moderator where appropriate. At intake, reliable psychiatric diagnoses (using SCID-5-CV and SCID-5-PD by American Psychiatric Association), childhood trauma (Childhood Trauma Questionnaire; (96); French version(97)), rejection sensitivity (Rejection Sensitivity Questionnaire-Adult; (98)) and the level of intelligence (NART reading test, French version (99)) will be assessed.
Manipulation checks will be introduced by using self-report questionnaires of arousal (Self-Assessment Manikin; SAM; (100)), self-esteem (State Self-Esteem Scale; SSES;(101)) and vividness (Vividness of Visual Imagery Questionnaire; VVIQ; (102, 103)).
Research Procedure
After the patient makes a request for treatment at the Department of Psychiatry, University of Lausanne, for problems related with BPD, the patient meets with a researcher who explains to him/her the study and informs him/her about the randomization and the assessment schedule . Upon approval, patients are block-randomized (by 10) and attributed either to a) a 4-month GPM or b) 4-month TAU (see Fig. 1). Patients are paid CHF 70.- for full participation in a assessment point.
Data analytic plan
For the behavioral assessments, we will conduct intent-to-treat and completer analyses for all variables (hypothesis 1: outcome, defined as residual gains at discharge). We will use multilevel modeling (104) where appropriate, for hypotheses 2 and 3 (global change and treatment response). For hypothesis 4 (mediation), we will conduct a mediation analysis for both potential mechanisms of change (105). Raw and composite scores for outcome and all mechanisms of change indexes will be used, by controlling for the corresponding fMRI data from the same assessment point. Composite scores involve for EP combining pre-post change in contemptuousness with change in neuronal regions of interest for individualized self-criticism and for SCP combining pre-post change in CCRT pervasiveness with change in neuronal regions of interest for theory of mind.
Therapist effects will be controlled for in the three-level HLM (106). All indexes (i.e., behavioral and fMRI) of the patient groups (N = 80) at intake will be rigorously compared with the indexes found for the healthy control (N = 20); we expect systematic between-group differences, in the context of a control analysis. Statistical treatment packages HLM7 and SPSS23 will be used for the analyses of the behavioral indexes.
For the fMRI assessments, we will use the methodology of blood-oxygen-level-dependant (BOLD) imaging followed by standard data processing and statistical analysis in the framework of SPM12. The fMRI data will be acquired on a Siemens Prisma 3T (64-channel head coil using a 2D EPI sequence). The acquisition parameters will be as follows: 3 × 3 × 3 mm3: TE = 30 ms, slice TR = 66 ms, 30 slices, flip angle = 90°. The structural MRI data consists of T1-weighted MPRAGE images (TR = 2000 ms; TI = 920 ms; α = 9°; BW = 250 Hz / pixel; readout in inferior-superior direction; FoV = 256 × 232 mm; 176 slices) at 1 mm resolution. All data pre-processing will be performed using the Statistical Parametric Mapping software (SPM12; Wellcome Trust Centre for Neuroimaging, shttp://www.fil.ion.ucl.ac.uk/spm/) running under Matlab 7.13 (The MathWorks, Inc., Natick, MA, USA). EPI images will be realigned to the subject’s average image across trials (corrected for spatial distortions using the SPM fieldmap tools). The parameters of registration to standardized MNI space will be calculated on the anatomical image and the default settings of the “unified segmentation” framework followed by the diffeomorphic registration algorithm DARTEL (107, 108). The spatial registration parameters will be applied to the functional time-series co-registered to the corresponding individual’s anatomical scan. Prior to statistical analysis, we will apply a spatial smoothing with a Gaussian kernel of 8 mm full-width-at-half-maximum. All statistical analyses will be performed using the default settings in SPM12. The statistical analysis at subject-specific level will be performed using the General Linear Model (GLM) after convolving the event onsets with a canonical hemodynamic response function (109). Both time-points will be modelled as two separate sessions within the design matrices. For the EP task, we will calculate at the subject level the interaction between WORDS (self-critical vs standard negative words; the non-words and the negative words will be excluded from the analyses, but used as control variables to ensure cognitive appropriateness) and TIME (time point 1 vs time point 2) using symbols as baseline. For the SCP task, the subject-level differential t-contrast will test the interaction between ToM, PUN and time point (the control stimuli will be excluded from the data analysis, but will serve as control for cognitive appropriateness). For both tasks, we will use a one-sample t-test along with the outcomes and arousal changes associated with treatment as regressors for the group-level analyses. The differential contrasts at the group level will test the positive and negative correlation between the interaction at the subject-specific level and BOLD signal changes. Where appropriate, we will control for the corresponding behavioural data from the same assessment point.
Monitoring and ethics
According to the accepted Data Management Plan, data will be entered into RedCap on a secured space on the University server . This program allows full accountability of data management, and should problems arise, standard procedures are in place. As such, data safety is guaranteed by the system’s security check and no formal data monitoring committee is requested (confidentiality ). The competent Ethics Committee can request audits at any moment in time and the primary investigator and his team will follow its instructions. Audit may include to disclose, in a restricted fashion and only if deemed necessary, personal data related to participants to the Ethics Committee . The same Data Management Plan outlines procedures in case of adverse events in the context of the trial which includes provision, if needed of post-trial care in case of harm .
The trial was approved by the competent ethics committee (see above) and potential amendments will have to be approved by the same, and be communicated to publishing journals. Informed consent will be obtained before the inclusion process can start by the competent researcher. For the fMRI part, this procedure involves the complete information on the safety recommendations in this environment. Participants’ personal information will be stored at a separate, locked, place at the Department of Psychiatry. No personal information will be revealed neither before, during nor after the trial (except for the case of audit by the competent Ethics Committee; confidentiality ; plans to give access to protocol, participant level-data and statistical code are described in the Data Management Plan ).
Publications of the results will be encouraged to all relevant groups (i.e., scientific publication, communication at conferences, communication with stakeholders, patients and families, ).
Feasibility: Results From The Pilot Study
We demonstrated feasibility of the pre-post design (50). We demonstrated for N = 8 medication-free right-handed female patients with BPD undergoing a 10-session psychiatric treatment that hypotheses 1 and 2 may be confirmed (due to the small sample size, the analyses for hypotheses 3 and 4 were not tested). The behavioral pre-post treatment outcome effect sizes ranged between d = 0.41 (for OQ-45) and d = 0.51 (for BSL-23). We observed arousal increase within session of the two-chair dialogue (d = 0.36), paralleled by a large decrease in peak arousal between pre- and post-treatment (d = 0.80). In the EP task, we demonstrated treatment associated trends for neural activity reduction in the associative parts of putamen when exposed to the individual’s own self-critical words. The exposure to ToM stimuli revealed trends for treatment related neural activity modulation in the OFC, ACC and NAcc, and the medio-dorsal nucleus of the thalamus. Neural activity (i.e., in the precuneus, left amygdala) related with the behavioral changes in arousal, but remained independent from outcome, whereas change in arousal was related with symptom reduction. Feasibility of the trial and relevance of the pre-post hypotheses are therefore demonstrated.