Quantifying person-level brain network functioning to facilitate clinical translation
2017; 7: e1248
Brain activation during fear extinction predicts exposure success
DEPRESSION AND ANXIETY
2017; 34 (3): 257-266
Although advances in neuroimaging have yielded insights into the intrinsic organization of human brain networks and their relevance to psychiatric and neurological disorders, there has been no translation of these insights into clinical practice. One necessary step toward clinical translation is identifying a summary metric of network function that is reproducible, reliable, and has known normative data, analogous to normed neuropsychological tests. Our aim was therefore to establish the proof of principle for such a metric, focusing on the default mode network (DMN). We compared three candidate summary metrics: global clustering coefficient, characteristic path length, and average connectivity. Across three samples totaling 322 healthy, mostly Caucasian adults, average connectivity performed best, with good internal consistency (Cronbach's ?=0.69-0.70) and adequate eight-week test-retest reliability (intra-class coefficient=0.62 in a subsample N=65). We therefore present normative data for average connectivity of the DMN and its sub-networks. These proof of principle results are an important first step for the translation of neuroimaging to clinical practice. Ultimately, a normed summary metric will allow a single patient's DMN function to be quantified and interpreted relative to normative peers.
View details for PubMedID 29039851
Single-Subject Anxiety Treatment Outcome Prediction using Functional Neuroimaging
2014; 39 (5): 1254?61
Exposure therapy, a gold-standard treatment for anxiety disorders, is assumed to work via extinction learning, but this has never been tested. Anxious individuals demonstrate extinction learning deficits, likely related to less ventromedial prefrontal cortex (vmPFC) and more amygdala activation, but the relationship between these deficits and exposure outcome is unknown. We tested whether anxious individuals who demonstrate better extinction learning report greater anxiety reduction following brief exposure.Twenty-four adults with public speaking anxiety completed (1) functional magnetic resonance imaging during a conditioning paradigm, (2) a speech exposure session, and (3) anxiety questionnaires before and two weeks postexposure. Extinction learning was assessed by comparing ratings to a conditioned stimulus (neutral image) that was previously paired with an aversive noise against a stimulus that had never been paired. Robust regression analyses examined whether brain activation during extinction learning predicted anxiety reduction two weeks postexposure.On average, the conditioning paradigm resulted in acquisition and extinction effects on stimulus ratings, and the exposure session resulted in reduced anxiety two weeks post-exposure. Consistent with our hypothesis, individuals with better extinction learning (less negative stimulus ratings), greater activation in vmPFC, and less activation in amygdala, insula, and periaqueductal gray reported greater anxiety reduction two weeks postexposure.To our knowledge, this is the first time that the theoretical link between extinction learning and exposure outcome has been demonstrated. Future work should examine whether extinction learning can be used as a prognostic test to determine who is most likely to benefit from exposure therapy.
View details for DOI 10.1002/da.22583
View details for PubMedID 27921340
Using neuroimaging to predict relapse in stimulant dependence: A comparison of linear and machine learning models.
2019; 21: 101676
The possibility of individualized treatment prediction has profound implications for the development of personalized interventions for patients with anxiety disorders. Here we utilize random forest classification and pre-treatment functional magnetic resonance imaging (fMRI) data from individuals with generalized anxiety disorder (GAD) and panic disorder (PD) to generate individual subject treatment outcome predictions. Before cognitive behavioral therapy (CBT), 48 adults (25 GAD and 23 PD) reduced (via cognitive reappraisal) or maintained their emotional responses to negative images during fMRI scanning. CBT responder status was predicted using activations from 70 anatomically defined regions. The final random forest model included 10 predictors contributing most to classification accuracy. A similar analysis was conducted using the clinical and demographic variables. Activations in the hippocampus during maintenance and anterior insula, superior temporal, supramarginal, and superior frontal gyri during reappraisal were among the best predictors, with greater activation in responders than non-responders. The final fMRI-based model yielded 79% accuracy, with good sensitivity (0.86), specificity (0.68), and positive and negative likelihood ratios (2.73, 0.20). Clinical and demographic variables yielded poorer accuracy (69%), sensitivity (0.79), specificity (0.53), and likelihood ratios (1.67, 0.39). This is the first use of random forest models to predict treatment outcome from pre-treatment neuroimaging data in psychiatry. Together, random forest models and fMRI can provide single-subject predictions with good test characteristics. Moreover, activation patterns are consistent with the notion that greater activation in cortico-limbic circuitry predicts better CBT response in GAD and PD.
View details for PubMedID 24270731
View details for PubMedCentralID PMC3957121
Intrinsic functional connectivity predicts remission on antidepressants: a randomized controlled trial to identify clinically applicable imaging biomarkers
2018; 8: 57
OBJECTIVE: Relapse rates are consistently high for stimulant user disorders. In order to obtain prognostic information about individuals in treatment, machine learning models have been applied to neuroimaging and clinical data. Yet few efforts have been made to test these models in independent samples or show that they can outperform linear models. In this exploratory study, we examine whether machine learning models relative to linear models provide greater predictive accuracy and less overfitting.METHOD: This longitudinal study included 63 methamphetamine-dependent (training sample) and 29 cocaine-dependent (test sample) individuals who completed an MRI scan during residential treatment. Linear and machine learning models predicting relapse at a one-year follow up that were previously developed in the methamphetamine-dependent sample using neuroimaging and clinical variables were applied to the cocaine-dependent sample. Receiver operating characteristic analysis was used to assess performance using area under the curve (AUC) as the primary outcome.RESULTS: Twelve individuals in the cocaine-dependent sample remained abstinent, and 17 relapsed. The linear models produced more accurate prediction in the training sample than the machine learning models but showed reduced performance in the testing sample, with AUC decreasing by 0.18. The machine learning models produced similar predictive performance in the training and test samples, with AUC changing by 0.03. In the test sample, neither the linear nor the machine learning model predicted relapse at rates above chance.CONCLUSIONS: Although machine learning algorithms may have advantages, in this study neither model's performance was sufficient to be clinically useful. In order to improve predictive models, stronger predictor variables and larger samples are needed.
View details for PubMedID 30665102
Computational Psychiatry: New Perspectives on Mental Illness (Book Review)
AMERICAN JOURNAL OF PSYCHIATRY
2017; 174 (7): 698-+
Individualized relapse prediction: Personality measures and striatal and insular activity during reward-processing robustly predict relapse (vol 152, pg 93, 2015)
DRUG AND ALCOHOL DEPENDENCE
2017; 175: 255
Neural Predictors of Initiating Alcohol Use During Adolescence
AMERICAN JOURNAL OF PSYCHIATRY
2017; 174 (2): 172-185
Default mode network (DMN) dysfunction (particularly within the anterior cingulate cortex (ACC) and medial prefrontal cortex (mPFC)) has been implicated in major depressive disorder (MDD); however, its contribution to treatment outcome has not been clearly established. Here we tested the role of DMN functional connectivity as a general and differential biomarker for predicting treatment outcomes in a large, unmedicated adult sample with MDD. Seventy-five MDD outpatients completed fMRI scans before and 8 weeks after randomization to escitalopram, sertraline, or venlafaxine-XR. A whole-brain voxel-wise t-test identified profiles of pretreatment intrinsic functional connectivity that distinguished patients who were subsequently classified as remitters or non-remitters at follow-up. Connectivity was seeded in the PCC, an important node of the DMN. We further characterized differences between remitters, non-remitters, and 31 healthy controls and characterized changes pretreatment to posttreatment. Remitters were distinguished from non-remitters by relatively intact connectivity between the PCC and ACC/mPFC, not distinguishable from healthy controls, while non-remitters showed relative hypo-connectivity. In validation analyses, we demonstrate that PCC-ACC/mPFC connectivity predicts remission status with >80% cross-validated accuracy. In analyses testing whether intrinsic connectivity differentially relates to outcomes for a specific type of antidepressant, interaction models did not survive the corrected threshold. Our findings demonstrate that the overall capacity to remit on commonly used antidepressants may depend on intact organization of intrinsic functional connectivity between PCC and ACC/mPFC prior to treatment. The findings highlight the potential utility of functional scans for advancing a more precise approach to tailoring antidepressant treatment choices.
View details for PubMedID 29507282
Individualized relapse prediction: Personality measures and striatal and insular activity during reward-processing robustly predict relapse
DRUG AND ALCOHOL DEPENDENCE
2015; 152: 93?101
Underage drinking is widely recognized as a leading public health and social problem for adolescents in the United States. Being able to identify at-risk children before they initiate heavy alcohol use could have immense clinical and public health implications; however, few investigations have explored individual-level precursors of adolescent substance use. This prospective investigation used machine learning with demographic, neurocognitive, and neuroimaging data in substance-naive adolescents to predict alcohol use initiation by age 18.Participants (N=137) were healthy substance-naive adolescents (ages 12-14) who underwent neuropsychological testing and structural and functional magnetic resonance imaging (sMRI and fMRI), and then were followed annually. By age 18, 70 youths (51%) initiated moderate to heavy alcohol use, and 67 remained nonusers. Random forest classification models generated individual alcohol use outcome predictions based on demographic, neuropsychological, sMRI, and fMRI data.The final random forest model was 74% accurate, with good sensitivity (74%) and specificity (73%). The model contained 34 predictors contributing to alcohol use by age 18, including several demographic and behavioral factors (being male, higher socioeconomic status, early dating, more externalizing behaviors, positive alcohol expectancies), worse executive functioning, and thinner cortices and less brain activation in diffusely distributed regions of the brain. Inclusion of neuropsychological, sMRI, and fMRI data significantly increased the prediction accuracy of the model.The results provide evidence that multimodal neuroimaging data, as well as neuropsychological testing, can be used to generate predictions of future behaviors such as adolescent alcohol use with significantly better accuracy than demographic information alone.
View details for DOI 10.1176/appi.ajp.2016.15121587
View details for Web of Science ID 000396662200015
TOWARD THE APPLICATION OF FUNCTIONAL NEUROIMAGING TO INDIVIDUALIZED TREATMENT FOR ANXIETY AND DEPRESSION
DEPRESSION AND ANXIETY
2014; 31 (11): 920-933
Nearly half of individuals with substance use disorders relapse in the year after treatment. A diagnostic tool to help clinicians make decisions regarding treatment does not exist for psychiatric conditions. Identifying individuals with high risk for relapse to substance use following abstinence has profound clinical consequences. This study aimed to develop neuroimaging as a robust tool to predict relapse.68 methamphetamine-dependent adults (15 female) were recruited from 28-day inpatient treatment. During treatment, participants completed a functional MRI scan that examined brain activation during reward processing. Patients were followed 1 year later to assess abstinence. We examined brain activation during reward processing between relapsing and abstaining individuals and employed three random forest prediction models (clinical and personality measures, neuroimaging measures, a combined model) to generate predictions for each participant regarding their relapse likelihood.18 individuals relapsed. There were significant group by reward-size interactions for neural activation in the left insula and right striatum for rewards. Abstaining individuals showed increased activation for large, risky relative to small, safe rewards, whereas relapsing individuals failed to show differential activation between reward types. All three random forest models yielded good test characteristics such that a positive test for relapse yielded a likelihood ratio 2.63, whereas a negative test had a likelihood ratio of 0.48.These findings suggest that neuroimaging can be developed in combination with other measures as an instrument to predict relapse, advancing tools providers can use to make decisions about individualized treatment of substance use disorders.
View details for PubMedID 25977206
View details for PubMedCentralID PMC4458160
Prefrontal dysfunction during emotion regulation in generalized anxiety and panic disorders
2013; 43 (7): 1475-1486
Functional neuroimaging has led to significant gains in understanding the biological bases of anxiety and depressive disorders. However, the ability of functional neuroimaging to directly impact clinical practice is unclear. One important method by which neuroimaging could impact clinical care is to generate single patient level predictions that can guide clinical decision-making. The present review summarizes published functional neuroimaging studies of predictors of medication or psychotherapy outcome in major depressive disorder, obsessive-compulsive disorder (OCD), posttraumatic stress disorder, generalized anxiety disorder, panic disorder, and social anxiety disorder. In major depressive disorder and OCD, there is converging evidence of specific brain circuitry that has both been implicated in the disordered state itself, and where pretreatment activation levels have been predictive of treatment response. Specifically, in major depressive disorder, greater pretreatment ventral and pregenual anterior cingulate cortex (ACC) activation may predict better antidepressant medication outcome but poorer psychotherapy outcome. In OCD, activation in the ACC and orbitofrontal cortex has been inversely associated with pharmacological treatment response. In other anxiety disorders, research in this area is just beginning, with the ACC potentially implicated. However, the question of whether these results can directly translate to clinical practice remains open. In order to achieve the goal of single patient level prediction and individualized treatment, future research should strive to establish replicable models with good predictive performance and clear incremental validity.
View details for DOI 10.1002/da.22299
View details for PubMedID 25407582
Selective effects of social anxiety, anxiety sensitivity, and negative affectivity on the neural bases of emotional face processing
2012; 59 (2): 1879-1887
The mechanisms that contribute to emotion dysregulation in anxiety disorders are not well understood. Two common disorders, generalized anxiety disorder (GAD) and panic disorder (PD), were examined to test the hypothesis that both disorders are characterized by hypo-activation in prefrontal cortex (PFC) during emotion regulation. A competing hypothesis that GAD in particular is characterized by PFC hyper-activation during emotion regulation (reflecting overactive top-down control) was also evaluated. Method Twenty-two medication-free healthy control (HC), 23 GAD, and 18 PD participants underwent functional magnetic resonance imaging (fMRI) during a task that required them to reappraise (i.e. reduce) or maintain emotional responses to negative images.GAD participants reported the least reappraisal use in daily life, and reappraisal use was inversely associated with anxiety severity and functional impairment in these participants. During fMRI, HCs demonstrated greater activation during both reappraisal and maintenance than either GAD or PD participants (who did not differ) in brain areas important for emotion regulation (e.g. dorsolateral and dorsomedial PFC). Furthermore, across all anxious participants, activation during reappraisal in dorsolateral and dorsomedial PFC was inversely associated with anxiety severity and functional impairment.Emotion dysregulation in GAD and PD may be the consequence of PFC hypo-activation during emotion regulation, consistent with insufficient top-down control. The relationship between PFC hypo-activation and functional impairment suggests that the failure to engage PFC during emotion regulation may be part of the critical transition from dispositionally high anxiety to an anxiety disorder.
View details for DOI 10.1017/S0033291712002383
View details for Web of Science ID 000320449300013
View details for PubMedID 23111120
View details for PubMedCentralID PMC4308620
Assessing Emotion Regulation in Social Anxiety Disorder: The Emotion Regulation Interview
JOURNAL OF PSYCHOPATHOLOGY AND BEHAVIORAL ASSESSMENT
2011; 33 (3): 346-354
Neural Mechanisms of Cognitive Reappraisal of Negative Self-Beliefs in Social Anxiety Disorder
2009; 66 (12): 1091-1099
Individuals with high anxiety show heightened neural activation in affective processing regions, including the amygdala and insula. Activations have been shown to be correlated with anxiety severity, but although anxiety is a heterogeneous state, prior studies have not systematically disentangled whether neural activity in affective processing circuitry is uniquely related to specific domains of anxiety. Forty-five young adults were tested on an emotional face processing task during functional magnetic resonance imaging. Participants completed the Social Interactional Anxiety Scale, Anxiety Sensitivity Index, and Spielberger Trait Anxiety Inventory. Using a robust multiple regression approach, we examined the effects of social anxiety, anxiety sensitivity, and trait anxiety (which overlapped with depressive symptoms, and can therefore be considered a measure of negative affectivity) on activation in insula, amygdala, and medial prefrontal cortex, in response to emotional faces. Adjusting for negative affectivity and anxiety sensitivity, social anxiety was associated with activity in left amygdala, right insula, and subgenual anterior cingulate across all emotional faces. When comparing negative and positive faces directly, greater negative affectivity was uniquely associated with less activity to positive faces in left amygdala, left anterior insula, and dorsal anterior cingulate. The current findings support the hypothesis that hyperactivity in brain areas during general emotional face processing is predominantly a function of social anxiety. In comparison, hypoactivity to positively valenced faces was predominantly associated with negative affectivity. Implications for the understanding of emotion processing in anxiety are discussed.
View details for DOI 10.1016/j.neuroimage.2011.08.074
View details for PubMedID 21920442
Social anxiety disorder (SAD) is characterized by distorted negative self-beliefs (NSBs), which are thought to enhance emotional reactivity, interfere with emotion regulation, and undermine social functioning. Cognitive reappraisal is a type of emotion regulation used to alter NSBs, with the goal of modulating emotional reactivity. Despite its relevance, little is known about the neural bases and temporal features of cognitive reappraisal in patients with SAD.Twenty-seven patients with SAD and 27 healthy control subjects (HCs) were trained to react and to implement cognitive reappraisal to downregulate negative emotional reactivity to NSBs, while undergoing functional magnetic resonance imaging and providing ratings of negative emotion experience.Behaviorally, compared with HCs, patients with SAD reported greater negative emotion both while reacting to and reappraising NSBs. However, when cued, participants in both groups were able to use cognitive reappraisal to decrease negative emotion. Neurally, reacting to NSBs resulted in early amygdala response in both groups. Reappraising NSBs resulted in greater early cognitive control, language, and visual processing in HCs but greater late cognitive control, visceral, and visual processing in patients with SAD. Functional connectivity analysis during reappraisal identified more regulatory regions inversely related to left amygdala in HCs than in patients with SAD. Reappraisal-related brain regions that differentiated patients and control subjects were associated with negative emotion ratings and cognitive reappraisal self-efficacy.Findings regarding cognitive reappraisal suggest neural timing, connectivity, and brain-behavioral associations specific to patients with SAD and elucidate neural mechanisms that might serve as biomarkers of interventions for SAD.
View details for DOI 10.1016/j.biopsych.2009.07.014
View details for Web of Science ID 000272599500004
View details for PubMedID 19717138
View details for PubMedCentralID PMC2788040