Brain markers predicting response to cognitive‐behavioral therapy for social anxiety disorder: an independent replication of Whitfield-Gabrieli et al. 2015

Abstract

Predictive brain markers promise a number of important scientific, clinical, and societal applications. Over 600 predictive brain markers have been described in published reports, but very few have been tested in independent replication attempts. Here, we conducted an independent replication of a previously published marker predicting treatment response to cognitive-behavioral therapy for social anxiety disorder from patterns of resting-state fMRI amygdala connectivity1. The replication attempt was conducted in an existing dataset similar to the dataset used in the original report, by a team of independent investigators in consultation with the original authors. The precise model described in the original report positively predicted treatment outcomes in the replication dataset, but with marginal statistical significance, permutation test p = 0.1. The effect size was substantially smaller in the replication dataset, with the model explaining 2% of the variance in treatment outcomes, as compared to 21% in the original report. Several lines of evidence, including the current replication attempt, suggest that features of amygdala function or structure may be able to predict treatment response in anxiety disorders. However, predictive models that explain a substantial amount of variance in independent datasets will be needed for scientific and clinical applications.

Brain Imaging Biomarkers for Chronic Pain

“…Harris et al ( 196 ), showed that treatment with pregabalin but not placebo altered insula chemistry and connectivity in patients with fibromyalgia but neither treatment was accompanied by a significant change in clinical pain rating. Most recently, Ashar et al used fMRI as “an objective correlate of treatment effects” in a clinical trial (NCT0394148) testing pain reprocessing therapy vs. placebo in patients with CLBP ( 197 ) and reported decreased anterior middle cingulate cortex activity in response to evoked clinical pain and increased anterior insula to somatosensory cortex connectivity with pain reprocessing therapy more than with placebo treatment. In addition, Reckzeigel et al ( 198 ) recently used brain biomarkers to assess the risk of transition from sub-acute to chronic pain in sub-acute low-back pain patients entering a pharmacological clinical trial (NCT01951105) aimed at preventing the transition to CLBP.…”

Informing the treatment of social anxiety disorder with computational and neuroimaging data

“…Santos et al ., 2019 ; Fig. 1C ), this method also demonstrates low replicability (Ashar et al ., 2021 ) reflecting issues concerning the reliability of fMRI measures (Noble et al ., 2019 ) and the heterogeneity of mental health disorders (Forbes et al ., 2023 ). As one example, significant within-group differences observed in precuneus and amygdala rsFC among individuals with SAD, in the absence of any group-level difference with healthy controls (Mizzi et al ., 2024 ), further necessitates an approach accounting for patient heterogeneity when employing a predictive (neural) model (Talmon et al ., 2021 ).…”

Brain Functional Connectivity and Anatomical Features as Predictors of Cognitive Behavioral Therapy Outcome for Anxiety in Youths

“…Although promising results exist, CPM is still understudied, as is rsFC for anxiety disorders more broadly (Zugman et al 2023). Only two studies used rsFC analyzed with methods different from CPM, to predict treatment outcome in anxiety (Whitfield-Gabrieli et al 2016; Ashar et al 2021). Both studies examined adults, and the predictive model failed to replicate.…”