Effect of D-Cycloserine on the Effect of Concentrated Exposure and Response Prevention in Difficult-to-Treat Obsessive-Compulsive Disorder

Key Points Question Does D-cycloserine potentiate the effect of concentrated exposure and response prevention in difficult-to-treat obsessive-compulsive disorder? Findings In this randomized clinical trial of 163 participants, D-cycloserine did not significantly affect treatment outcomes. Most patients responded to the concentrated exposure and response prevention treatment, and nearly 50% were recovered at 1-year follow-up. Meaning In this study, concentrated exposure and response prevention treatment was effective for patients with difficult-to-treat obsessive-compulsive disorder, but adding D-cycloserine did not potentiate the treatment.


Introduction
Obsessive-compulsive disorder (OCD) is a debilitating psychiatric disorder with a lifetime prevalence of 1% to 2%, 1,2 often with onset in childhood. 3 Exposure and response prevention (ERP) and antidepressants are effective treatments, 4,5 but response is mixed, indicating a need to develop more effective strategies. Based on experimental animal studies demonstrating that D-cycloserine (DCS), a partial N-methyl-D-aspartate receptor agonist, 6,7 facilitates extinction learning, a number of human clinical trials have assessed whether ERP is augmented by DCS. While some early publications were promising, 8,9 more recent meta-analyses show only small augmentation of DCS vs placebo (effect size, 0.25) 10 and others no significant difference, [11][12][13] including no significant difference in several studies involving patients with OCD. 10,11,[14][15][16] A study by Andersson et al 17 concluded that DCS did not augment the effects of cognitive behavioral therapy, but found that antidepressants may interact with DCS to block its facilitating effect on fear extinction.
The DCS studies included in meta-analyses are based on samples in which a treatment response would be expected for 62% to 68% of the sample. 4 Potentiation by DCS may be more evident in patients who have a documented history of being difficult to treat, ie, those who have not responded to ERP or who have responded but then relapsed. These patients also constitute the group most in need of alternative treatment approaches. One study 18 reported that DCS accelerated the rate of recovery in children with difficult-to-treat OCD, suggesting that DCS might also be helpful for adults with difficult-to-treat OCD.
The present study targets patients with OCD who have a documented history of nonresponse to or relapse following ERP treatment and investigates whether DCS potentiates the effect of concentrated ERP treatment immediately after treatment and at 12-month follow-up. We hypothesized that patients receiving DCS would demonstrate significantly greater reductions in OCD symptoms relative to participants in the placebo group.

Methods
Norwegian health authorities have established 15 specialized adult OCD teams, yielding national coverage. Nine teams recruited participants for the study. Data collection, management, training of therapists, and organization of treatment were conducted from the Bergen site. Details regarding the clinical training procedure are described elsewhere. 19,20 A total of 8 group leaders and 64 therapists participated. The study was approved by the regional committees for medical and health research ethics in Norway, and all participants provided written consent. The trial protocol is available in Supplement 1. The study followed the Consolidated Standards of Reporting Trials (CONSORT) guideline.

JAMA Network Open | Psychiatry
Effect of D-Cycloserine on Concentrated Exposure and Response Prevention in Difficult-to-Treat OCD

Design and Participants
All patients received concentrated ERP treatment delivered during 4 consecutive days in groups of 3 to 6 patients with a 1:1 ratio between patients and therapists. The 2 middle days were used for exposure treatment (eFigure 1 in Supplement 2). Effectiveness studies in routine clinical care [19][20][21][22] as well as a randomized clinical trial 23 have demonstrated that 90% of patients with OCD respond to ERP treatment delivered in this format and that 70% are recovered at 4-year follow-up, 24 based on the international consensus criteria. 25 DCS was administered both days of exposure treatment. Given that research on optimal dosage has been inconclusive, 10,13 100 mg and 250 mg dosages were evaluated. Participants were stratified by use of antidepressants. 17 Thus, the study used a triple-masked, 3-group, placebo-controlled design, in which patients within each stratum were randomized to 100 mg DCS, 250 mg DCS, or placebo in a 2:2:1 ratio for an intended sample of 160. Randomization in blocks of 5 was done using an online tool before the first patient was included in the study and concealed from all patients, therapists, and independent assessors. Due to the group treatment format, the actual sample size was 67 of 163 (41.1%) in the 250 mg group, 65 (39.9%) in the 100 mg group, and 31 (19.0%) in the placebo group. The trial was announced through media and on the websites of the Norwegian OCD association and the OCD teams. Inclusion lasted from January 2016 to August 2017.

Inclusion Criteria
We included patients who met Diagnostic and Statistical Manual of Mental Disorders (Fifth Edition) (DSM-5) criteria for OCD; were able to be treated as outpatients; were aged at least 18 years; were fluent in Norwegian; and had either responded to and relapsed following or not responded to prior ERP treatment, consisting of at least 6 sessions of ERP. Response to earlier ERP was defined by an at least 35% reduction and a posttreatment Yale-Brown Obsessive Compulsive Scale (Y-BOCS) score of 15 or lower; relapse was defined by an at least 35% increase in Y-BOCS score from posttreatment, a Y-BOCS score of 16 or more, and a Clinical Global Impression (CGI) improvement score of 6 (ie, "much worse") or higher. 25 Nonresponders were defined as those with a reduction in Y-BOCS scores from pretreatment to posttreatment of less than 35% and a Y-BOCS score of at least 16 after treatment. A minimum of 4 weeks since treatment ended was required.

Exclusion Criteria
Patients who had ongoing substance abuse and/or dependence; had bipolar disorder or psychosis; had active suicidal ideation or plans; had not receiving a stable dosage of antidepressants for at least 12 weeks or were not willing to receive a stable dosage during the 4 intervention days; were unwilling to refrain from anxiety-reducing substances during the 2 days of exposure; had an intellectual disability; and were living more than 1 hour by car or train from the treatment location were excluded.
Exclusion criteria related to the DCS were pregnancy or breastfeeding, kidney impairment, hypersensitivity to DCS, porphyria, and epilepsy.

Adherence and Competence
All group sessions and therapist meetings were videotaped. Also, each group had a trained therapist who observed and evaluated whether the group was conducted in accordance with the protocol.
No deviances from the protocol were reported for any of the groups. Two experts on concentrated ERP who had not participated in the given group independently scored all videotapes for adherence and competency using a 3-point scale. With 1 exception, both experts rated all groups as adherent and competent.

Independent Assessors
All Y-BOCS and Structured Clinical Interview for DSM-5 (SCID-5) were conducted by specially trained and independent assessors and were audiotaped. A second assessor rated 20% of the taped interviews. The obtained κ coefficient of diagnostic agreement was excellent (κ = 0.92). The same procedure was used for the Y-BOCS interviews, and the interrater reliability of total score was excellent (intraclass correlation coefficient, 0.94).

DCS
Each patient received 1 capsule of DCS (100 mg or 250 mg) or placebo each of the 2 days of exposure. DCS and placebo were prepared in identical capsules by a research pharmacy. The participants received written information about the medication and a phone number for questions or to report adverse events. Some prior trials have indicated that DCS might potentiate negative experiences when exposure is brief and not followed by a reduction in anxiety. 26 To minimize this risk, the first capsule was taken when the patient had gained experience with the procedure, ie, at lunchtime on day 2.
At posttreatment, assessors were asked to guess which dosage of medication the patients had received and also to indicate how certain they were of their guesses. The result showed no correspondence between the actual group and guesses (χ 2 4 = 2.62; P = .62). Certainty was rated from 0 to 10, and the mean (SD) score was 4.0 (3.0). Patients were asked the same questions and the results showed a mean (SD) certainty of 0.3 (0.9), and no correspondence between the actual group and their guesses (χ 2 4 = 2.45; P = .65).

Diagnostic Procedure
All patients were screened for inclusion using Y-BOCS and Mini International Neuropsychiatric Interview (MINI). 27 Patients receiving a preliminary OCD diagnosis after the MINI had a diagnostic interview using the SCID-5. 28 The SCID-5 was carried out by a team of masked independent assessors who had undergone extensive training in the procedure.

Primary Outcome Measures
Patients were assessed pretreatment, posttreatment, and at 3-month and 12-month follow-ups. The Y-BOCS 29 was the primary outcome measure. In addition to mean scores on Y-BOCS, clinical improvement was evaluated using a modified version (not including the CGI) of the international consensus criteria, which defines response as a reduction of at least 35% of pretreatment Y-BOCS score and remission as the response criterion plus a posttreatment Y-BOCS score of 12 points or lower. 25 Recovery was defined as in remission at 1-year follow-up. We also used the criteria by Jacobson and Truax 30 for clinically significant change; the results were very similar to those obtained with the international consensus criteria (eTable in Supplement 2). Finally, we reported change in diagnostic status (DSM-5) as assessed by SCID-5 at 3-month and 12-month follow-ups.

Statistical Analysis
A meta-analysis 37 found a moderate effect size for ERP with DCS for heterogeneous samples of patients with OCD. Given the inclusion of only cognitive behavioral therapy nonresponders or patients who had relapsed after ERP, we anticipated a larger effect size compared with previous ERP studies. To have 80% power to detect a moderate effect size (d = 0.50) at an α of .05, a total of 160 patients needed to be included (64 in each of the DCS groups and 32 in the placebo group). We compared Y-BOCS scores between the placebo and DCS (100 mg and 250 mg) groups from pretreatment to posttreatment and 2 follow-ups using mixed-effects regression models. 37 The dropout at the different points of assessment was very low (Figure). Following the principle of intention to treat, all participants were included in the analyses, irrespective of missing data at any measurement point. 38 Time was treated as a categorical variable because we did not expect a strictly linear effect of time. Between-group differences in Y-BOCS were assessed by including fixed effects for treatment group, time, and the time × treatment group interaction. Bonferroni confidence intervals were used for all analyses.
The mixed model included random intercepts for all participants. To take the potential clustering effect of group leader into account, we also included a random effect for group leader.
Participants were also stratified according to use of antidepressants. To examine a possible effect of current use of antidepressants, a 3-way interaction among time, treatment condition, and current use of antidepressants was included in the analysis. Mixed models also allow the use of realistic variance and correlation patterns to achieve more efficient statistical inference and, therefore, greater statistical power. 38 In this study, an unstructured covariance structure was used. This choice was informed by comparing plausible covariance structures using goodness-of-fit statistics (deviance, Akaike information criterion, and Bayesian information criterion). 38 Intervention (withingroup) effect sizes were estimated using Glass Δ, with pretreatment SD as denominator. The manipulation in intervention studies will often affect the SD as well as the mean; therefore Glass Δ is

Pretreatment Characteristics
The total sample mean (SD) age was 34

Number of Groups and Sites
There were a total of 36 groups with group sizes ranging from 3 to 6 patients. Nine clinics were involved in the trial, of which 4 (Bergen, Oslo, Kristiansand, and Trondheim) treated 125 patients (76.7%). There were no significant differences in Y-BOCS scores between the 4 clinics at the 4 points of assessment.

Treatment Effect of DCS vs Placebo
Results on the primary and secondary outcome measures are presented in Table 2. To test the hypothesis that DCS (regardless of dosage) enhances ERP, we specified a contrast that compared the 2 DCS groups (250 mg and 100 mg) with the placebo group and interacted this with time. This interaction was not statistically significant (χ 2 3 = 6.23; P = .10). Treating the 2 DCS groups as independent also resulted in a statistically nonsignificant time × treatment group interaction (χ 2 6 = 7.59; P = .27). Effect sizes for the primary and secondary outcome measures are presented in Table 3.
There was a significant main effect of time (χ 2 3 = 776.26; P < .001). The reduction in Y-BOCS score from pretreatment to posttreatment was significant for all treatment groups (placebo, - 13 Table 4). There was a significant reduction in symptoms at 12 months, and within-group effect sizes ranged from 3.01 (95% CI, 2.38-3.63) for the group receiving 250 mg DCS to 3.49 (95% CI, 2.78-4.18) for the group receiving 100 mg DCS (all P < .001). However, there was no significant effect of treatment group compared with placebo in obsessive-compulsive symptoms at posttreatment (250 . Unlike the 2 DCS groups, the GAD-7 score in the placebo group significantly increased from posttreatment to the 3-month follow-up (2.30; 95% CI, 0.14 to 4.47). At the 3-month follow-up, the mean GAD-7 score for the placebo group was also significantly higher than that of the 250 mg DCS group (mean difference, 3.11; 95% CI, 0.53 to 5.67) but not that of the 100 mg DCS group. There were no statistically significant differences between the groups at the 12-month follow-up.

Client Satisfaction and Adverse Events
Participants were generally satisfied with the treatment as indicated by a mean score of 28.9 (SD,   17 we did not find any significant effect of ongoing treatment with antidepressants on the effect of DCS.

Limitations
This study has limitations. The lack of a DCS effect in the current study was not due to an overall inferior effect of the ERP treatment. Compared with means across 5 previous randomized clinical trials 8,9,14,16,24  In contrast to previous studies, the maskedness of the assessors was tested in the current study; their guesses of which treatment each patient had received was random. It is also noteworthy that only 1 patient dropped out of therapy and only 1 other did not complete the posttreatment assessment. Thus, selective attrition was not a problem in this study. Furthermore, the sample primarily consisted of women. Also, there was a slight difference between groups in pretreatment Y-BOCS scores.
The nonsignificant findings correspond with previous findings. 10,11,[14][15][16] The study was powered to detect an effect size of 0.50 because a smaller effect size would have limited clinical significance.
This is especially the case given that the targeted group was patients with difficult-to-treat OCD. A 2019 meta-analysis 40 suggested that more DCS doses (up to 9) and administering DCS more than 60 minutes before exposure were associated with better outcomes. However, the concentrated format used in the present study is not compatible with administering more than 2 doses. Previous research has also indicated that a possible effect occurs during treatment or immediately after treatment. 8,9,18 Given our concentrated treatment format and having the first follow-up visit after 3 months, the theory of accelerated improvement was not tested. Given the nonsignificant results, we suggest that future research should test the suggestions brought forth by Rosenfield et al. 40

Conclusions
In this randomized clinical trial, DCS did not potentiate the effect of concentrated ERP for patients with difficult-to-treat OCD. However, concentrated ERP treatment was associated with improvement in symptoms of OCD, anxiety, and depression.