Post by Elisa Guma
What's the science?
Major depression is a chronic and disabling mood disorder, for which the primary treatment is traditionally antidepressant medications. These are effective for a subset of individuals, however, for many they do not afford much improvement over placebo treatment. Given this heterogeneity, it is thought that an individuals’ neurobiological characteristics might predict their response to treatment. This week in Nature Human Behaviour, Fonzo and colleagues investigated whether an individual’s response to or regulation of emotional conflict and associated neural response could predict treatment outcome to an antidepressant, sertraline, compared to placebo in a double-blind trial.
How did they do it?
The authors used functional magnetic resonance imaging (MRI) data on an emotional conflict task from a large double-blind trial, Establishing Moderators and Biosignatures of Antidepressant Response for Clinical Care for Depression (EMBARC). In the trial, participants were randomized into an 8-week course of sertraline (an antidepressant) or placebo, and depressive symptoms were measured each week. The authors chose to focus their study on emotional conflict, as its regulation is critical to wellbeing, it engages cognitive and emotion related brain circuitry, is often dysregulated in depression, and is thought to be of relevance to the efficacy of antidepressant treatment. The task performed while undergoing functional MRI involved the presentation of an emotional face with either a fearful or happy expression, with an overlaid emotional word (i.e. fear or happy). Participants were instructed to identify the facial emotion while trying to ignore the emotional word. Stimuli were either presented as congruent (i.e. face and word match), or incongruent (emotional conflict). First, the brain regions that were more active during this task were identified, as well as the relationship between their activity and task performance. Next, the authors built a prediction model (using relevance vector machines — a type of machine learning model) to assess whether the neural response associated with emotional conflict regulation could predict treatment outcome.
What did they find?
In keeping with other large clinical trials, the sertraline treated group had a slightly larger improvement in symptoms than the placebo group. During the task, the incongruent trials (emotional conflict trials) resulted in a slower reaction time. The neural response to this conflict was identified as a conflict response network and included activation of the dorsomedial and dorsolateral prefrontal cortices, ventrolateral prefrontal cortex and anterior insula, and deactivation of the ventromedial prefrontal cortex and anterior medial prefrontal cortex as well as the posterior cingulate, precuneus, hippocampus, and parahippocampal gyri. The authors found that better emotional conflict regulation (i.e. mitigation of the conflict effect on conflict trials that were preceded by another conflict trial) was positively associated with greater dampening of the conflict response network activation.
In addition, the authors found that individuals who were more successful at dampening their emotional conflict response network during conflict regulation (i.e. adapting to interference when the face and word do not match) also had more symptom improvement due to sertraline rather than placebo. Interestingly, the prediction model the authors built was able to reliably predict how treatment changed depressive symptoms when the model was trained (built) using data from the group receiving sertraline, but it did not work when trained on the placebo group. This suggests that the model developed on sertraline outcome reflects a sertraline-specific signal that explains a meaningful effect of treatment, separate from the placebo response. The model was not related to or affected by depression diagnosis or clinical severity.
What's the impact?
The authors were able to identify a neural trait sensitive to medication response within a broader clinical diagnosis of depression. This trait was related to the brain’s ability to adapt its response to emotional conflict and was a better predictor of treatment outcome than either the clinical measures or the behavioural response alone. Future studies could incorporate a broader range of brain imaging modalities and behavioural assessments, and assess generalizability to other antidepressant medications. These findings highlight the heterogeneity in treatment response within a clinically defined population and suggest that individual neurobiological characteristics could predict treatment response.
Fonzo et al. Brain regulation of emotional conflict predicts antidepressant treatment response for depression. Nature Human Behaviour (2019). Access the original scientific publication here.