Exposure-based treatments are – on average – highly effective and recommended as first-line treatment for anxiety disorders. Yet, not all patients benefit equally well. Investigating the mechanisms of action and predictors of treatment outcomes has become central to the field of psychotherapy research, as it may establish the basis for an optimized, personalized treatment selection and design.
Towards this goal, uncovering the mechanisms of pathological avoidance behavior, which may prevent patients from engaging in exposure, has gained growing interest. Further, associative learning processes, including fear extinction and fear generalization, have been theorized to moderate outcomes of exposure-based treatments – with preliminary empirical support. Complementing investigations on avoidance and fear learning, predictive-modelling studies may support the a-priori identification of patients at risk for treatment-non-response. This symposium will cover recent advances regarding treatment-relevant (bio-)behavioral learning processes and present potential moderators and predictors of treatment outcomes in anxiety disorders.
First, Andre Pittig will present his work on low-cost and costly avoidance and their association with symptom severity in mixed anxiety disorders. Second, Jan Richter will focus on extinction learning in anxiety disorders. Based on data from the Germany-wide research network Protect-AD, he will discuss the relation between implicit and explicit markers of fear extinction and highlight its associations with exposure therapy outcomes. Third, Kati Roesmann will present behavioral and magnetoencephalographic evidence for associations between pre-treatment overgeneralization of fear and non-response to exposure therapy in spider phobia. The symposium will conclude with a presentation by Kevin Hilbert, who will discuss predictors for individual treatment success in anxiety and related disorders based on evidence from machine learning approaches.
11:30 Uhr
Low-cost and costly avoidance in mixed anxiety disorders
Andre Pittig, Würzburg (Germany)
Details anzeigen
Autor:in:
Andre Pittig, Würzburg (Germany)
Pathological avoidance is a transdiagnostic characteristic of anxiety disorders. Avoidance conditioning reemerged as translational model to examine mechanisms and treatment of avoidance. However, its validity for anxiety disorders remains unclear.
This study tested for altered avoidance in patients with anxiety disorders compared to matched controls (N=80) using an instrumental conditioning task assessing low-cost avoidance (avoiding a single aversive outcome) and costly avoidance (avoidance conflicted with gaining rewards). Autonomic arousal and threat expectancy were assessed as psychophysiological and cognitive indicators of conditioned fear. Associations with dimensional symptom severity were examined.
Patients and controls showed frequent low-cost avoidance without group differences. Controls subsequently inhibited avoidance to gain rewards, which was amplified when aversive outcomes discontinued. In contrast, patients failed to reduce avoidance when aversive and positive outcomes competed (elevated costly avoidance) and showed limited reduction when aversive outcomes discontinued (persistent costly avoidance). Symptom severity was associated with costly avoidance and its persistence, but not with low-cost avoidance. Interestingly, elevated costly avoidance was not linked to higher conditioned fear in patients. Moreover, individual data revealed a close to binary distribution of costly avoidance: Some patients showed persistent avoidance, whereas others showed little to no avoidance. Persistent versus low avoiders did not differ in other task-related factors, response to gains and losses in absence of threat, sociodemographic data, or clinical characteristics.
Findings suggest that anxious psychopathology is associated with a deficit to inhibit avoidance in presence of competing positive outcomes. This offers novel perspectives for research on mechanisms and treatment of anxiety disorders.
11:40 Uhr
Considering the complexity of fear extinction – about the relation between implicit and explicit markers and its associations to exposure therapy outcomes
Jan Richter, Greifswald (Germany)
11:50 Uhr
Associations between overgeneralization of fear and non-response to virtual reality exposure therapy in spider phobia
Kati Roesmann, Siegen (Germany)
Details anzeigen
Autor:in:
Kati Roesmann, Siegen (Germany)
Fear generalization – and its pathological form, fear overgeneralization – are considered crucial factors in the maintenance of anxiety disorders. Here, we investigated whether pre-treatment fear generalization in adult spider phobic patients moderates treatment response to exposure therapy (ET). 90 patients with spider phobia (SP) completed a One-Session Virtual Reality ET, a clinical and a magnetoencephalography (MEG) assessment before, and a clinical assessment after therapy. Based on self-reported symptom reductions in the Spider Phobia Questionnaire, patients were categorized as either responders ( > 30% reduction) or non-responders. The MEG-assessment consisted of baseline, conditioning, and subsequent generalization phases. In the conditioning phases, aversive unconditioned stimuli (US) were either paired or never paired with differently tilted Gabor gratings (CS+, CS-). In the subsequent generalization phases, fear ratings, US expectancy ratings and event-related fields to CS+, CS- and seven different generalization (GS) stimuli that ranged on a perceptual continuum from CS+ to CS- were obtained. Non-responders compared to responders showed behavioral overgeneralization indicated by more linear generalization gradients in fear ratings. Analyses of MEG source estimations revealed that linear generalization gradients in frontal clusters also differentiated between (later) non-responders and responders. While stronger (inhibitory) frontal activations to safety-signaling CS- and GS compared to CS+ declined over time in non-responders, responders maintained these activations at early ( < 300ms) and late processing stages. Results provide initial evidence that pre-treatment differences of behavioral and neural markers of fear generalization may hold predictive information regarding later responses to behavioral exposure. Our findings demonstrate the relevance of inhibitory learning functions and their spatio-temporal neural reflections in this interplay.
12:00 Uhr
Predicting treatment outcome in anxiety and related disorders with machine learning
Kevin Hilbert, Berlin (Germany)
Details anzeigen
Autor:in:
Kevin Hilbert, Berlin (Germany)
Although psychotherapy, and specifically exposure therapy, are effective treatments for anxiety disorders, a substantial proportion of patients still does not benefit. Precision psychotherapy aims to improve treatment outcomes by tailoring treatments to patient characteristics. To this end, identifying patients who are at-risk for exposure non-response is an important first step. Here, machine learning approaches may be crucial due to their ability to integrate a large number of different, potentially interacting predictor variables into a single outcome prediction for the individual patient. In two studies, we aimed to predict individual exposure-therapy treatment outcomes and to identify predictors of non-response. The first study investigated a naturalistic sample of n=533 OCD patients in a specialized outpatient clinic, while the second study examined n = 174 spider phobia patients in a highly standardized single-session virtual reality exposure. Importantly, the latter study was conducted at two independent study sites enabling a test of prediction generalizability from one site to the other. In both studies, individual treatment outcomes were predicted with comparable, moderate accuracy. However, when attempting to generalize a model from one site to the other, prediction accuracy dropped considerably. In line with the literature, baseline severity scores were crucial to prediction. Overall, both studies demonstrate the potential of machine learning for individual outcome prediction, but also indicate that additional informative predictors are needed to improve prediction performances to clinical utility.