Supplementary Appendix
“Electrical stimulation in the bed nucleus of the stria terminalis
alleviates severe obsessive-compulsive disorder”
Laura Luyten, Sarah Hendrickx, Simon Raymaekers, Loes Gabriëls & Bart Nuttin
Table of contents
Supplementary Methods2
Clinical study2
Patient selection2
Surgical procedure2
Initial electrical stimulation3
Double-blind randomized crossover design3
Follow-up3
Patient evaluation3
Neuroimaging4
Statistical analyses4
Psychiatric outcome (crossover trial, after 4 years of follow-up and at last follow-up)4
Neuropsychological outcome (crossover trial)4
Stimulation target versus outcome at last follow-up5
Neuroanatomical analyses5
Localization of active electrode contacts5
Determination of stimulated volume5
Supplementary Results6
Serious adverse, device-related and adverse events6
Serious adverse events6
Device-related adverse events6
Adverse events6
Table S1: Serious adverse events7
Table S2: Complete list of adverse events8
Conclusion10
Stimulation parameters and electrode polarity – crossover trial11
Table S3: Stimulation characteristics crossover ON phase11
Neuropsychological tests – crossover trial12
Results12
Conclusion12
Table S4: Neuropsychological test scores13
Qualitative analysis – crossover trial14
Introduction & Method14
Results14
Conclusion15
Fig. S1: Summary of the qualitative analysis15
Table S5: Exhaustive list of the reported observations16
Active contacts during crossover, optimal open-label period and at last follow-up18
Optimal open-label period18
Comparison of stimulation targets during crossover and optimal open-label period18
Comparison of stimulation targets during optimal open-label period and at last follow-up18
Conclusion18
Patient data included in previously published papers19
Supplementary References20
Supplementary Methods
Clinical study
Patient selection
Between 1998 and 2010, 24 patients were selected by a Committee for Neurosurgery for Psychiatric Disorders according to strict selection criteria[1, 2]. Inclusion criteria were a principal diagnosis of OCD of disabling severity, based upon a structured clinical interview for the Diagnostic and Statistical Manual of Mental Disorders: DSM-IV [3], with a Yale-Brown Obsessive-Compulsive Scale (Y-BOCS) score of at least 30/40 and a Global Assessment of Functioning (GAF) score of 45 or less [3-5]. This level of impairment should have persisted for a minimum of 5 years, despite adequate trials (except in case of intolerance) with two selective serotonin reuptake inhibitors and clomipramine, augmentation strategies (i.e. antipsychotics), and cognitive behavioral therapy. Patients had to be between 18 and 60 years of age, and able to understand and comply with instructions and provide their own written informed consent to be included in the study. Patients were excluded in case of a current or past psychotic disorder, any clinically significant disorder or medical illness affecting brain function or structure (other than motor tics or Gilles de la Tourette syndrome), or current or unstably remitted substance abuse. Comorbid anxiety or depressive disorder was not an exclusion criterion. Clinical evaluations and surgeries were performed at the University Hospitals of Antwerp (psychiatric assessment until 2007) and Leuven (psychiatric assessment from 2007 onwards and all surgeries), Belgium. During a period of at least 4-6 weeks before surgery, medication was tapered off to a minimum and no psychotherapy was administered.
The study protocol was reviewed and approved by the University Hospitals’ Ethics Committees and is in accordance with the Declaration of Helsinki of 1975 (Rev. 1983).
Surgical procedure
Two quadripolar electrodes (Models 3887, 3387 or 3391; Medtronic, Inc., Minneapolis, MN, USA) were stereotactically implanted into the bilateral anterior limbs of the internal capsule (ALIC), similar to the targets used for anterior capsulotomy [6]. The most ventral contact (contact 0) was implantedin the grey matter ventral to ALIC and the other contacts (1-2-3) were placed in ALIC. Preliminary analyses correlating electrode position and Y-BOCS scores indicated a better outcome with a slightly more posterior target.For subsequent patients, this was formalized in a new protocol that was approved by the Ethics Committee (Clinical Trial NCT01985815). The optimized target was more posterior, ventral and medial, with at least one contact (usually contact 0) in the bed nucleus of the stria terminalis (BST). Taking into account vascularization, both electrodes were implanted symmetrically through precoronal burr holes. Note that Patient 5 received additional electrodes bilaterally aimed at the mediodorsal thalamus, on account of its role in the CSTC circuit and because its connections with the prefrontal cortex are interrupted with classical anterior capsulotomy[7, 8]. Prophylactic antibiotics were administered. Surgery was performed under general (n = 4) or local (n = 20) anesthesia, depending on the patient’s choice. Within two weeks after electrode implantation, bilateral implantable pulse generators (IPGs) (Itrel 2, Synergy, or Kinetra; Medtronic, Inc.) were implanted in the abdominal (n = 21) or subclavicular (n = 3) area and connected to the electrodes.
Supplementary AppendixLuyten et al.2015 – DOI: tbd1/21
Initial electrical stimulation
After implantation of the electrodes, initial stimulation parameters were explored by the neurosurgeon and psychiatrist, using an external programming device. Several combinations of the four electrode contacts, using different frequencies, amplitudes and pulse widths were assessed, based on acute (within 5 minutes) mood improvement or reduction of anxiety or obsessive thoughts, while the patient was unaware of the stimulation parameters. Contact combinations with the lowest threshold for these favorable effects were used. After 1-2 weeks, the IPGs were implanted. During the next weeks or months, stimulation parameters were further optimized. This time-consuming procedure evaluated virtually all contact combinations, high and low frequencies and pulse widths. Amplitudes were increased slowly or quickly, sometimes up to 10.5 V, which is the maximal voltage delivered by the IPG. The aim of this optimization period was to determine stimulation parameters resulting in satisfactory alleviation of OCD symptoms, without adverse effects. The parameters should have a relatively stable effect, in order to be of use during the crossover phase of this clinical study.
Double-blind randomized crossover design
After the initial optimization period, patients entered the crossover design, with 3 months of stimulation ON and 3 months of stimulation OFF. This part of the study was double-blind (patient and evaluating psychiatrist were not informed about the stimulation condition) and randomized (patients were assigned to an ON-OFF or OFF-ON order) [9]. The optimized stimulation parameters and electrode polarity were maintained throughout the crossover ON phase, while in the OFF phase no stimulation (0 V) was delivered. An ‘escape procedure’ was foreseen if a crossover phase led to intolerable suffering. For instance, it was possible that during the optimization period or potentially a preceding stimulation ON phase, patients experienced clear therapeutic effects and that returning to a status without stimulation (OFF phase) would be unbearable. If this was the case, the patient was thoroughly evaluated by the blinded psychiatrist, who then decided if the crossover phase could be prematurely ended. If the shortened phase was the first crossover period, the neurosurgeon changed the parameters and started the next phase, without unblinding the patient and psychiatrist.
After completing both crossover phases, the patient and psychiatrist were unblinded, and the patient could choose to be continuously stimulated. Fine-tuning of stimulation is a continuing process, and, if indicated, parameters (i.e., amplitude, pulse width, frequency, choice of cathodes) were adapted during follow-up.
Follow-up
The first patients have now been followed up for 16 years, while the last patients have had a follow-up of 4 years. In this paper, we will report on the stimulation effects for all patients, 4 years after implantation and at last follow-up (May 2014).
In addition, we determined the optimal open-label period (and corresponding stimulation parameters), i.e. a period of at least 2 months during which the patient experienced the lowest Y-BOCS scores since implantation. Detailed descriptions and analyses of the patients’ long-term clinical evolution are ongoing and will be reported separately.
Patient evaluation
At several time points, patients were evaluated using standardized psychiatric and neuropsychological questionnaires and tests.
The primary outcome measure, assessing the treatment effect on obsessions and compulsions, was the psychiatrist-rated Y-BOCS [4, 5], which was obtained 2 weeks before surgery, during the ON and OFF phases of the crossover, and approximately 4 years after implantation. In addition, it was regularly rated during psychiatric consultations. A 35% decrease in Y-BOCS was defined as the responder criterion [1]. At the same time, we evaluated the patients’ depressive and anxiety symptoms (Hamilton Depression Rating Scale (HAM-D) and Hamilton Anxiety Rating Scale (HAM-A)) and their overall functioning (Global Assessment of Functioning (GAF)) [3, 10, 11].
In addition to these psychiatric assessments, several neuropsychological evaluations were performed before surgery and during the ON and OFF phases: Complex Figure Test of Rey, Audio Verbal Learning Test, Wisconsin Card Sorting Test, Word Fluency test, Raven Standard progressive matrices, STROOP and Trail Making Tests A en B [12-16].
Besides these standardized, quantitative evaluations, patients were regularly seen by the neurosurgeon and psychiatrist, resulting in qualitative reports of how the patient was doing, including favorable and potential side effects of the treatment. These lists of (serious) adverse events were classified as being device-related, surgery-related, (probably) (not) stimulation-induced or unrelated to DBS.
Neuroimaging
Structural magnetic resonance imaging (MRI), computed tomography (CT) and/or functional MRI (fMRI) scans were obtained before and after surgery. In addition, part of the patients underwent 18F-fluorodeoxyglucose positron emission tomography (PET), before surgery and during the crossover study [1]. The nuclear imaging results have been reported elsewhere[17].
Statistical analyses
Psychiatric outcome (crossover trial, after 4 years of follow-up and at last follow-up)
Crossover trial (n = 17). Percentage improvement was calculated as (OFF minus ON)/OFF for Y-BOCS, HAM-A and HAM-D scores. Improvement in GAF score was calculated as ON minus OFF. Preoperative, ON and OFF scores were compared using a Friedman ANOVA (p < .05), followed by Wilcoxon matched pairs tests with correction for multiple testing (Bonferroni: .05/3, p < .017).
Four-year follow-up (n = 18) and last follow-up (n = 24). Percentage improvement was calculated as (preoperative minus follow-up score)/preoperative score for Y-BOCS, HAM-A and HAM-D data. Improvement in GAF score was calculated as follow-up minus preoperative score. Preoperative and follow-up scores were compared using a Wilcoxon test (p < .05). Correlation between the Y-BOCS score during the crossover ON phase and at follow-up was calculated using a Spearman coefficient (p < .05).
Neuropsychological outcome (crossover trial)
Test scores (7 tests, resulting in a total of 22 subscores) were analyzed using repeated measures one-way ANOVA (preoperative vs ON vs OFF) with Tukey's posthoc tests. If the sphericity assumption was violated (Mauchly's test p < .05), a Greenhouse-Geisser correction was applied. If the normality and/or equal variance assumptions were violated, data were analyzed using a non-parametric Friedman ANOVA, followed by Wilcoxon tests with correction for multiple testing (Bonferroni: .05/3, p < .017). For exploratory purposes, the significance level of the ANOVA main effects was set at .05. Using a more stringent criterion, we also checked significance of the main effects after correction for multiple testing (Bonferroni: .05/22 subtests, p < .002). Analyses were conducted on data from 16 patients (neuropsychological tests were not carried out in one patient who completed the crossover design).
Stimulation target versus outcome at last follow-up
Using the predefined criterion of at least 35% Y-BOCS improvement to define a satisfactory effect of stimulation, we investigated if one of both brain targets (i.e., ALIC or BST) resulted in a better outcome than the other at last follow-up, in a 2x2 contingency table with a Fisher's exact test (p < .05). Note that this trial was not conceived as a study to compare two targets. Nevertheless, the relative variation of the active contacts’ location does allow us to evaluate the target-outcome relation with this posthoc analysis.
Neuroanatomical analyses
Localization of active electrode cathodes
It is generally assumed that stimulation effects mainly take place at the negative contacts (cathodes), while clinically relevant effects near the positive anode(s) are negligible [18-20]. Accordingly, if only one contact is used, it is programmed as a cathode, while the case of the pulse generator is positive. In DBS for movement disorders, it is common to use this type of monopolar stimulation, as it results in radial current diffusion, with relatively low stimulation intensities compared to other electrode configurations [21]. Other options include bipolar (one cathode and one anode on the same lead) or multipolar (more than one cathode and/or anode on the same lead) stimulation. Bipolar stimulation creates a more focused stimulated area, mainly near the cathode, which can be useful to limit side effects elicited by more diffuse monopolar stimulation. In addition, it can be indicated to have several adjacent cathodes, to create a larger, elongated stimulated volume [18, 21].
To investigate the locus of action of DBS in more depth, we indicated the center of the active contacts (cathodes) on Mai’s brain atlas [22]. Using pre- and post-operative MR and CT scans, we localized the cathodes for each patient during crossover, the optimal open-label period and at last follow-up. If available, we merged pre-operative MR and post-operative CT scans using Medtronic FrameLinkTM Software (Medtronic Inc., Minneapolis, MN, USA). For the other patients, post-operative MR scans were used. All digital images were manually reformatted along the AC-PC plane. Finally, contact positions were determined by two observers (LL & BN) and transferred to the atlas plates.
Determination of stimulated volume
In addition to the localization of the active contacts, we also estimated the extent of brain tissue influenced by stimulation. The stimulated volume is largely spherical or ellipsoid, and depends on several factors, e.g., stimulation parameters, electrode impedance and size, localization of the contact in gray or white matter and mono-, bi- or multipolar stimulation [23]. However, it can be assumed that, for monopolar stimulation and stimulation with non-adjacent cathodes, about 1 mm per volt (starting from the center of the negative contact) will be influenced, up till a voltage of about 4 V. With higher voltages, the radius will be proportionally smaller. See Table S3 for an overview of the estimated radii for each contact, based upon [24-28]. Adjacent cathodes produce a broader, more elliptical field of stimulation, with a radius perpendicular to the lead of about 1.2 mm per volt and a major axis extending about 1 mm per volt from the cathodes. For bipolar stimulation, the field spreading from the cathode has a radius of about 0.6 mm per volt.
We preferred a conservative and approximate calculation of the stimulated tissue, rather than attempting to model the volume in a more detailed way. First, because the existing complex models are primarily based on DBS targets for Parkinson’s disease, which differ in their anatomical properties from our target [29]. Second, because the additional clinical value of detailed estimations is most likely limited for our purposes.
Supplementary AppendixLuyten et al.2015 – DOI: tbd1/21
Supplementary Results
Serious adverse, device-related and adverse events
Serious adverse events
Table S1 shows the 25 serious adverse events that were recorded during 180 patient-years of follow-up in the full cohort of 24 patients.Note that one of the patients who experienced a tonic-clonic insult had a history of head and spine trauma, and was under influence of alcohol (5 units) and sleep deprivation. The other patient was not sleep-deprived, but was known to be a moderate alcohol abuser (2-3 units daily). One patient stated that she only experienced absences during stimulation, while another patient continued to have seizures after turning off stimulation. Another patient only developed symptoms after one electrode was replaced and implanted in a slightly posterior target.
In addition to the 25 serious events, we counted 23 hospitalizations (11 patients) which were illness-related and in some instances worsened by difficult family circumstances.
Device-related adverse events
There were several device failures and device-related adverse events during the long-term follow-up of our patients. Although most of them resulted in surgery (which is not surprising given that the devices are implanted), there were no life-threatening events, nor was there any permanent physical impairment.
In 2 patients, one of the leads was broken and therefore replaced. Three patients had faulty extensions (5 defects in total) which were revised. One patient had a malfunctioning stimulator which was replaced. In one patient, repositioning of the stimulator was indicated because of irritation and skin atrophy, due to weight loss. In 3 patients, stimulators were turned off (2 patients, on repeated occasions) or had been reset to factory settings (1 patient). The reasons for these unwanted changes in stimulation parameters were unknown, but in these cases, patients were reminded to avoid strong electric or magnetic fields (e.g. large stereo speakers with magnets, airport/security screening devices).
Adverse events
Apart from the abovementioned serious and device-related adverse events, patients reported many other complaints. In Table S2, these adverse events are listed and categorized as being surgery-related, stimulation-induced, probably (not) stimulation-induced or unrelated to DBS. To not overload the text with an enumeration of all complaints, we will only mention adverse events that were reported by at least 10 patients.
A common surgery-related complaint was an uncomfortable feeling around the extension cables (12/24 patients). Frequent stimulation-induced adverse events included fatigue (18/24), memory complaints (16/24, although this could not be objectified with neuropsychological tests in the crossover study, cf infra), disinhibition (12/24), increased assertiveness (12/24), logorrhea (10/24) and hyperactivity (10/24). Adverse events that were probably stimulation-induced, comprised weight gain (13/24), weight loss (10/24), insomnia (13/24), irritability (16/24), tension/nervousness (13/24) and family problems (13/24, probably partially stimulation-induced). Finally, there were some adverse events which were probably not stimulation-induced, e.g., depressed feelings (13/24), suicidal thoughts (12/24) and headache (11/24).
Supplementary AppendixLuyten, Gabriëls & Nuttin1/21