Optimizing exposure-based CBT for anxiety disorders via enhanced extinction: Design and methods of a multicentre randomized clinical trial

1.1 Exposure therapy for anxiety disorders

Exposure-based cognitive-behavioural therapies (CBTs) are currently the most effective interventions for the treatment of anxiety disorders [ADs; Bandelow et al., 2014; National Institute for Health and Clinical Excellence (NICE), 2011, 2013]. Several meta-analyses of randomized clinical trials (RCTs) (Butler, Chapman, Forman, & Beck, 2006; Hofmann & Smits, 2008; Norton & Price, 2007) have shown large and long-term effects for various ADs such as panic disorder and agoraphobia (Mitte, 2005; Sánchez-Meca, Rosa-Alcázar, Marín-Martínez, & Gómez-Conesa, 2010), social anxiety disorder (Mayo-Wilson et al., 2014), generalized anxiety disorder, or specific phobias (Wolitzky-Taylor, Horowitz, Powers, & Telch, 2008), with exposure being most often conceived as the central principle of change (Lohr, Lilienfeld, & Rosen, 2012; Neudeck & Wittchen, 2012). Despite the consistent evidence for the overall efficacy of exposure-based CBT, several research questions remain.

First, CBT is an umbrella term describing an intervention package that includes various components such as psychoeducation, cognitive restructuring, exposure-based interventions applied in situ, in sensu or in virtual reality contexts. These components are supposed to differentially target specific dysfunctional domains of the patient's symptomatology. Considerable variation in CBTs exists further by type of AD, as well as by form and intensity of cognitive-emotional and behavioural intervention components, making it difficult to pinpoint their respective effect. Thus, the core active ingredients of CBTs that promote change and the mechanisms involved in therapeutically induced change in specific ADs remain difficult to determine. Second, not all patients benefit equally well. Treatment attrition rates in AD psychotherapy studies range around 16–31% (Fernandez et al., 2015; Taylor, Abramowitz, & McKay, 2012). Of those who commence therapy, 40–47% fail to remit or relapse after successful treatment (Loerinc et al., 2015). Thus, the question arises whether augmenting central treatment components leads to better and more stable outcomes.

In the past, various mechanisms have been identified as possible mechanisms underlying exposure-based CBT, including within- and between-session habituation (Foa & Kozak, 1986), counterconditioning (Wolpe, 1995), or neurotrophic factors (Ströhle et al., 2010). More recent evidence points to extinction as the central process that underlies the reduction of fear (Milad & Quirk, 2012; Vervliet, Craske, & Hermans, 2013).

1.2 Experimental investigation of extinction

The term “extinction“ descends from paradigms of Pavlovian conditioning and is commonly used in various ways; (1) to describe the experimental procedure during which a conditioned stimulus (CS+) that has previously been paired with the unconditioned stimulus (US) is now presented without the US; (2) to describe the result of the procedure, the reduction of the fear response that can be observed even after days (long-term extinction); and (3) to describe the associative neuronal learning as well as memory process that underlies the reduction of the fear response. Here, the term extinction training will be used to describe the experimental procedure, fear reduction will be used to describe the decrease of the fear response, and extinction to describe the learning process that underlies the observed fear reduction (see Myers & Davis, 2007). Supporting the clinical relevance of extinction, studies indicated that patients with ADs show increased fear responses to the CS+ during extinction training compared to healthy controls (Duits et al., 2015; Lissek et al., 2005). This deficit in fear extinction may contribute to the intensity, generalization, and persistence of pathological anxiety. Deficits in extinction have also been associated with a subsequent onset of ADs (Lommen, Engelhard, Sijbrandij, van den Hout, & Hermans, 2013) and with non-response to exposure-based CBT (Fullana et al., 2016). Thus, impaired extinction may represent a central mechanism for both the development of ADs and their treatment (Craske et al., 2008). Understanding how extinction can be facilitated during exposure-based interventions might be crucial to optimize the effects of exposure therapy.

Phenomena such as reinstatement of fear – i.e. fast recovery of fear during mere presentations of the US after extinction – demonstrate that the CS–US associations are not simply forgotten (for a review see Vervliet et al., 2013). Instead, the individual has to actively learn that the feared cue is no longer followed by the aversive consequences. For example, a patient with agoraphobia who is afraid to faint (US) when using public transportation (CS+) holds the fear-evoking or excitatory association “public transport – fainting”. If, however, he uses a bus but does not faint, an additional inhibitory association (“bus – no fainting”) is formed, that will reduce the fear response when the CS memory in a bus) is activated. The initial excitatory association is not erased but rather modulated by the inhibitory association that is also stored in memory. Depending on strength and diversity of excitatory associations, fear memories may be retrieved even after successful extinction (Bouton, 2004). Thus, even after response to exposure-based treatment, fear and anxious responses may return. Strengthening the inhibitory non-fear associations may attenuate this return of fear and result in better long-term treatment outcome.

1.3 Clinical translation of extinction

Translating the extinction model to clinical practice, several behavioural strategies have been suggested to enhance exposure (Pittig, van den Berg, & Vervliet, 2016). Central to these strategies is to “put the patient's central concerns to a test”, because new, inhibitory learning is initiated when a prediction error occurs (Rescorla & Wagner, 1972): extinction is facilitated if the individuals' central concern is violated by the exposure experience. To that end, the patients' central concern needs to be precisely formulated prior to exposure and exercises individually designed to allow patients to test and disconfirm their predictions (Craske, Treanor, Conway, Zbozinek, & Vervliet, 2014). To increase the positive prediction error, safety signals, avoidance behaviours and other control strategies need to be abolished. Here, patients are instructed to renounce any behaviour or circumstance that might decrease their central concern (e.g. conducting exposure without the presence of an accompanying person, which is believed to prevent negative consequences). Furthermore, variation strategies such as variation of stimulus and context properties (Bouton, 2004) or compound or deepened extinction (Culver, Vervliet, & Craske, 2015) have been experimentally investigated (for a detailed description see Craske et al., 2014; Pittig et al., 2015).

Another enhancement strategy targets the timing or spacing between exposure exercises to augment extinction. While short-term extinction may be accelerated by shortening intervals between exposure sessions (Orinstein, Urcelay, & Miller, 2010; Tsao & Craske, 2000), this might attenuate long-term learning effects (Rowe & Craske, 1998). Long-term learning might instead be augmented by expanding the temporal spacing of exposure exercises (Bjork & Bjork, 2006). Thus, an intensified psychological treatment may combine temporally massed exposure followed by a gradual extension of intervals may result in optimal long-term learning and even in shorter treatment durations.

First, while there have been promising attempts to translate basic findings on extinction into clinical interventions, systematic clinical trials are still missing. For example, it remains unclear whether a temporally massed exposure scheme with spaced follow-up exercises yields stronger long-term reduction in clinical anxiety. Second, it is an open question whether interventions tailored to increase prediction error during exposure treatment will improve clinical outcome. Third, studies have often used selected mono-symptomatic clinical cases. It is thus not clear whether enhanced extinction is beneficial across all ADs and whether comorbid diagnoses will modulate these effects. Thus, there is a great potential to exploit advances in basic research to optimize extinction within the context of psychological treatment.

1.4 Aims of the research consortium

The research consortium “Providing Tools for Effective Care and Treatment of Anxiety Disorders” (PROTECT-AD) is a collaboration of anxiety researchers within the German National Research Network on Mental Disorders. Using an RCT and accompanying add-on projects, the consortium aims to test enhanced extinction as a mechanism of action in exposure-based CBT.

In IPI, intervals between treatment sessions are shortened and spaced exposure trials using context and stimulus variation are assigned during follow-up. The variation strategies are not specifically instructed in the comparison group, but are thoroughly documented in both groups in order to estimate their effects. The primary hypothesis of the RCT is: IPI will be superior to weekly exposure treatment resulting in faster, stronger, more stable and pervasive improvements in primary and secondary outcomes.

Associated research questions are: Is the clinical outcome associated with changes in everyday life (assessed via a combined actographic and ecological momentary assessment tool)? Is IPI associated with an enhanced positive prediction error (i.e. reduced individual expectancies of aversive outcomes) during and after exposure? Are the effects independent of type of diagnosis, comorbid depression and psychopharmacology? To compare extinction in different forms of ADs, patients with a primary DSM-5 (Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition) diagnosis of social anxiety disorder, panic disorder, agoraphobia and multiple specific phobias are included. Secondary disorders like generalized anxiety disorder, somatic symptom disorder, major depression and persistent depressive disorder are allowed.

In order to experimentally examine extinction and detect its moderators, specific add-on projects are implemented (see section 3). A further goal of the consortium is to disseminate exposure based treatments into clinical routine care. To that end, a transfer-oriented subproject is conducted, involving all major stakeholders in the healthcare system.

2.1 Research design

The RCT involves seven university centres in Germany. Based on the high evidence for the efficacy of exposure-based treatments in ADs, no waitlist control group was included. Instead, a comparative design with two active treatment arms was chosen: The IPI treatment arm is compared to treatment as usual (TAU).

IPI differs from TAU with regard to (a) the temporally massed structure of the exposure phase, (b) the instruction of spaced exposure trials including stimulus and context variation during follow-up. Temporally massed exposure in IPI is realized by providing the exposure module in two weeks with three sessions each compared to six weeks with one session each in TAU. Treatment dose, however, is equal in both groups (see section 2.4).

Study assessment points (see Figure 1) include baseline assessment (before inclusion), intermediate (after sessions 4 and 11), post (after session 12) and follow-up assessments (six months after end of treatment).

2.2 Sampling and eligibility

Target sample size to be included are N = 700 patients with primary DSM-5 anxiety disorders (Wittchen, Heinig, & Beesdo-Baum, 2014). Participants are primarily recruited from university outpatient clinics at each site. Participants are screened (age, primary anxiety complaints, availability and prior treatment, as described later) and invited for informed consent and a subsequent diagnostic baseline assessment to inspect eligibility criteria (see Table 1). Patients with acute anxiety-related pharmacotherapy [e.g. selective serotonin reuptake inhibitors (SSRIs), benzodiazepines] are excluded, whereas maintenance therapy (> three months and stable) is allowed if patients still meet inclusion criteria. Medication is kept stable during therapy. Concomitant psychotherapy is not allowed. If a patient decides to stop pharmacotherapy or a concomitant CBT prior to inclusion, a waiting period of two to three months is set. Medical relative contraindications involve conditions that impede thorough exposure, e.g. cardiovascular diseases, autoimmune diseases or pregnancy. As prior research shows that CBT for ADs also results in decreased negative affect and depression (Emmrich et al., 2012; Olatunji, Cisler, & Tolin, 2010), patients with comorbid depression are explicitly included.

Eligible patients are randomized to one of the two conditions. Randomization is performed using DatInf Randlist (version 1.2). A two-stage randomization procedure ensures that no single person is able to foresee group assignment. Blinding of patients and therapists is not feasible in psychotherapy studies.

Sample size was calculated based on the SIGH-A total score. For patients in IPI, a reduction of the primary outcome by 12 points at the end of treatment is expected, compared to a reduction by 10 points for patients in TAU. Using a power of 80%, a test significance level of alpha =5%, a standard deviation of 10 points and a one-sided t-test to detect differences between the IPI and TAU group, a total of 310 patients are needed per group, i.e. 620 patients are planned to be analysed (see Figure 2). Based on previous similar trials (Gloster et al., 2009) drop-out rates during treatment of 10 to 15% are expected.

2.3 Diagnostic domains and instruments

Primary efficacy endpoint is the clinician-rated SIGH-A (see Table 2 and Supporting Information for description of the instruments). Categorical diagnoses are assessed using the computerized version of the standardized clinical interview CIDI (DIA-X). A range of secondary outcomes was chosen to evaluate if IPI is associated with more pervasive changes in various domains. Proxy measures for extinction are subjective measures of within-session and between-session exposure effects (anxiety ratings, expectancy ratings of central concerns) and further assessments in the experimental add-on projects.

2.4 Treatment procedure

2.5 Study personnel

Therapies are realized by certified study therapists who are state-examined cognitive-behavioural psychotherapists or psychotherapy trainees in advanced stage of training. They receive a two-day practice-oriented training of the study manual. Therapists have to treat patients in both experimental groups to avoid therapist effects. Certification requires implementation and video recording of one IPI therapy, which is evaluated based on five core treatment sequences.

Study assessments are administered by certified clinical assessors, mostly psychology students. They receive a two-day treatment of clinical interviews (DIA-X/CIDI and SIGH-A), operational procedures and the web-based study database. Certification of assessors requires a video-recorded baseline assessment and a standardized SIGH-A rating.

2.6 Monitoring and quality assurance

Data management, including monitoring and validity checks, is effectuated by a central data management officer using the study software REDCap, a secure web application including an audit trail. All assessments and modifications are immediately accessible via web access. At each participating centre an assigned data manager is responsible for online transmission of data and site specific quality assurance. Supervision of data management according to GCP Guidelines (European Medicines Agency, 2002) lies with the Coordination Centre for Clinical Trials (KKS) Dresden.

The treatment manual is modular and highly structured by session goals and verbatim text suggestions for therapists. A comparable manual has been tested in RCTs and is widely used, treatment integrity and therapist compliance are high (Hauke et al., 2013). All therapy sessions are video recorded to monitor adherence by random video inspection (5% of sessions). Repeated violations lead to exclusion of the therapist.

All sites receive initial and regular site visits by study monitors to check protocol adherence. Protocol violations will be documented and lead to exclusion of patients, therapists or the centre.

2.7 Feasibility

To ensure feasibility of the trial, a pilot study was run using a preliminary version of the IPI manual. Forty-one patients [age: mean (M) = 33.2, standard deviation (SD) = 11.6] with all primary inclusion diagnoses and in many cases (41%) with depressive comorbidity were treated and the manual was adapted following patient's and therapist's feedback.

To assure local feasibility at each centre, 36 assessors and 56 therapists and supervisors were trained in the study procedures and therapy manual prior to recruitment. Prior to inclusion of the first patients, all centres conducted two complete IPI therapies including experimental components. Online monitoring of data entry was used to give centres direct feedback in case of protocol violations.

At the present moment (November 2016), 1306 patients have been screened and 204 have been included in the trial. Main reasons for exclusion are (a) no primary study diagnosis, (b) too little impairment and (c) no consent with IPI. One-hundred and six therapies have been completed. Two-hundred and thirty-eight violations of the study protocol have been registered. Of those, 56% concerned violations of the time scheme (e.g. deferral of sessions due to illness of patient or therapist) and 27% violations of the therapy protocol (e.g. session content not completely administered). In most cases (roughly 75%), effects on study therapies were estimated negligible. Twenty-one patients (10.3%) discontinued study therapy due to violations of the time scheme, non-compliance with the rationale or fast remission of symptoms, which lies within the anticipated range of dropout.

3.1 Psychophysiological and neural markers of extinction

The psychophysiological and neural add-on projects examine extinction and reinstatement using a two-day experimental extinction training protocol in both experimental groups. During day one, one of two neutral facial stimuli (CS+) embedded in either blue or yellow background colour is followed by an aversive electric US during six of 10 presentations while the other stimulus (CS˗) is never paired with the US. Since the focus of the study is on extinction we want to make sure that each patient indeed will acquire a fear response. Therefore, we give an instruction prior to the acquisition phase stating which of the two facial stimuli would occasionally be followed by the aversive US and which one will not. On day two, extinction and reinstatement are tested during functional magnetic resonance imaging (fMRI). The paradigm starts with one re-acquisition trial followed by an extinction training during which both stimuli are presented again 20 times each without any presentation of the US. After presentation of three aversive US without any CS, 10 extinction trials for each CS will be presented to test the effect of reinstatement. Neural correlates of fear extinction, reinstatement and emotion processing are examined focussing on amygdala, (para-)hippocampal and anterior cingulate cortex (ACC) function (Lueken et al., 2013; Sehlmeyer et al., 2011). The paradigm is repeated after psychotherapy, using two different faces to avoid re-acquisition. In addition, a paradigm on emotion processing; structural scans and a resting state examination are conducted in the fMRI scanner. To disentangle psychotherapy and memory effects, the paradigm is also applied in a healthy control group (n = 100). This is the first study investigating extinction and reinstatement via fMRI before and after exposure therapy.

Due to physical contraindications, lack of informed consent and other exclusion reasons a sample size of n = 300 patients is expected for the fMRI part of the study. Patients with fMRI-related contraindications will perform the paradigm on fear extinction and reinstatement on day two in the psychophysiological laboratory. In addition, these participants are asked to participate in a context conditioning experiment using a virtual reality paradigm on day two. In this paradigm, participants can enter one of two rooms from a corridor (ITI), and four aversive events (unpleasant scream) will occur in one room (context; CTX+) but not in the other (CTX˗). Immediately after acquisition, extinction of context anxiety will be investigated.

Associated exploratory research questions are: Do different indices of extinction vary with the diagnostic groups or age? Moreover, since this add-on project provides the link between findings in basic experimental research and clinical applications we will be able to test interactions between physiological and motor reflex fear read outs in a well-controlled experimental design with (epi)genetic variation in AD vulnerability genes and to investigate how these different moderators affect clinical outcome. Are amygdala and ACC activation during extinction training modulated by (epi)genetic variation in genes relevant for extinction prior to psychopharmacological treatment and depression comorbidity?

At present (November 2016), 156 fMRI measurements (102 pre- and 54 post-treatment) in patients have been performed. Thus, 50% of the recruited patients could be included in the fMRI add-on project. Additionally, 15 healthy control subjects have been investigated. Psychophysiological markers of extinction were investigated in 71 patients (34.80%) during day two prior to therapy (only patients who did not take part in the fMRI assessment); 36 patients participated in the psychophysiological laboratory at post-assessment.

3.2 (epi)genetic variation

Associated exploratory research questions are: Is there a moderator effect of diagnostic patterns and previous or accompanying pharmacotherapy? The identification of (epi)genetic markers – intertwined with psychophysiological and neural network markers – in the aetiology, course and comorbidity of ADs may aid in developing resilience-increasing preventive measures in high-risk groups. The definition of epigenetic signatures for core mechanisms of action of fear extinction (objective biomarker of treatment outcome) might contribute to the development of the targeted, personalized treatment of ADs.

At present (November 2016), 272 blood samples of 176 patients have been taken (176 at baseline, 83 post-therapy, 13 at follow-up). The rate of participation is thus 86.3%.

3.3 Ecological momentary assessment

The EMA during the exposure phase contains questions on the preparation, specific expectancies, emotions and evaluations of exposure exercises. Most importantly, expectancies are assessed multiple times before and after each exercise to evaluate the course of the expectancy of central concerns. This allows to examine if prediction error is constant once it has been experienced.