Cortical activation predicts posttraumatic improvement in youth treated with TF-CBT or CCT

1. Introduction

More than one fourth of youth within the United States will have been exposed to a traumatic event by late adolescence and more than a third will have experienced multiple traumatic events (Copeland et al., 2007; Copeland-Linder, 2008). Traumatic and adverse childhood experiences (TRACEs) are associated with many diverse outcomes and, if unaddressed, the impact may be enduring and last well into adulthood (Weems, Russell, Herringa, & Carrion., 2021). Childhood TRACEs increases the risk for a host of mental health conditions both in the short and long term including posttraumatic stress disorder (PTSD), anxiety, depression, attention deficit hyperactivity disorder (ADHD), suicidality, substance use, eating disorders, and psychosis (Alisic et al., 2014; Macdonald et al., 2010). Early developmental exposure to TRACEs also increases the risk for aggression, delinquent behavior, and violence in youth, these behaviors are closely linked to PTSD symptomatology (Aebi et al., 2017; Yoon et al., 2016). Childhood trauma may derail the normative trajectory of neurodevelopment resulting in impairments in executive functioning, memory, and learning (Carrion & Weems, 2017; Malarbi et al., 2017; Weems et al., 2021). Further, it can lead to structural and functional alterations in the brain and metabolic systems (Carrion & Weems, 2017; Shonkoff et al., 2012).

Many interventions have been developed to address the sequelae of childhood trauma ranging from individual, group, parent/family, school-based, and biological modalities. Core components of these interventions include trauma-informed child and parent assessment, stabilization and safety, psychoeducation about trauma responses, coping skills, emotional expression and regulation, cognitive processing, exposure to trauma reminders, consolidation of learned skills, and enhancement of self-efficacy and empowerment (Matlow, 2019). Evidence-based trauma-focused interventions are considered the first line of treatment for youth with posttraumatic stress (AACAP, 2010).

Of the evidence-based interventions for childhood trauma, trauma-focused cognitive behavioral therapy (TF-CBT) is the most widely researched (Cohen et al., 2017). Originally developed for use with sexual trauma, it has since been expanded to address a wide range of traumatic events including multiple traumatic experiences (Cohen et al., 2012). TF-CBT consists of nine PRACTICE components provided in three phases: 1) Stabalization Phase including psychoeducation, parenting, relaxation, affective modulation, and cognitive processing skills; 2) Trauma Narration and Processing phase, including trauma narration and processing; and 3) Integration Phase, including in-vivo mastery of traum reminders, conjoint child-parent sessions, and enhancing safety and future development. A wealth of studies demonstrates the efficacy of TF-CBT in reducing PTSD, depression, and other emotional and behavioral difficulties in traumatized youth (Cohen et al., 2017).

Cue-Centered Therapy (CCT) is a hybrid treatment combining elements of insight-oriented, CBT, psychodynamic, narrative, exposure, and family therapies (Carrion, 2016). It was designed to specifically address chronic trauma and the accumulation of stressors throughout the lifespan (i.e., allostatic load). The cue center focus, versus specific event focus, fosters intervention where multiple TRACEs are involved in precipitating emotional distress in youth. CCT includes caregiver participation in a few key sessions to minimize disruptions caused by barriers to caregiver attendance. CCT also aims to empower youth through facilitating insight into how traumatic cues develop through conditioning, and how these cues are linked to emotion, cognition, and behaviors, but also their physiology (Weissman et al., 2006). The four phases of CCT include: 1) psychoeducation and coping skills; 2) chronic traumatic stress history (i.e., trauma narratives and life timeline) and cognitive restructuring; 3) gradual exposure to trauma mimetic cues; 4) narrative retelling and consolidation of skills learned. A randomized controlled trial of youth with chronic trauma comparing CCT to a waitlist control group found that CCT was effective in reducing symptoms of PTSD, anxiety, and depression and improved overall functioning (Carrion et al., 2013). Caregivers in the CCT group also reported a significant reduction in anxiety. These gains were maintained at three-month follow-up.

Although interventions as TFCBT and CCT are effective there is variation in treatment response and evidence that different treatment modules may be relatively effective for different youth. Integrating the findings from neuroscience research among youth with TRACEs may help elucidate variability in treatment response (Weems & Neill, 2020). In particular, functional neuroimaging data suggest the potential of using brain biomarkers as objective assessment tools and even predictors of outcomes in youth with TRACEs (Carrion et al., 2008; Carrion et al., 2010; Garrett et al., 2012). Functional magnetic resonance imaging (fMRI) indicates that youth with TRACEs and PTSD may have differential activation and connectivity patterns in the amygdala, with concurrent involvement of the prefrontal cortex (i.e., medial, dorsolateral and ventrolateral prefrontal cortex; Herringa, 2017; Keding & Herringa, 2016; Wolf & Herringa, 2016). Frontocortical areas are vital for the appraisal of threatening stimuli and regulation of negative emotions (Etkin et al., 2011). Aberrant functioning (i.e., diminished PFC top-down control over amygdala) is thought to contribute to the heightened fear response in individuals with PTSD (Herringa et al., 2017; Koch et al., 2016; Sripada et al., 2012). The assessment of frontoamygdala activation and connectivity patterns allowed fMRI researchers to distinguish youth with TRACEs from those without (Garrett et al., 2012; Carrion et al., 2008; Keding & Herring,a 2016), and predict PTSD symptom improvement (Garrett et al., 2019; Cisler et al., 2016). Research suggests that establishing an understanding of the complex associations between functional activation in various brain areas and TRACEs among youth appears particularly promising for both research and clinical applications (Weems, Russell, Neill, & McCurdy, 2019; Weems et al., 2021; Weems & Neill, 2020). However, the high cost, motion sensitivity, and cumbersome environmental requirements of MRI hinder affordable and frequent assessment of these functional connectivity patterns.

In a first attempt to establish brain biomarkers of pediatric PTSD in an affordable and portable manner, Balters and colleagues (2021) examined functional activity among youth (n = 57) with TRACES and PTSD using a functional near-infrared spectroscopy (fNIRS) neuroimaging system. Hemodynamic responses were measured while participants engaged in a classic emotion expression task that included fearful and neutral face stimuli. A General Linear Model was applied to identify cortical activation associated with facial stimuli. Subsequently, a prediction model was established via a Support Vector Regression to determine whether PTSD severity could be predicted based on fNIRS-derived cortical response measures and individual demographic information. Results suggested increased activation in the dlPFC and vlPFC in response to negative emotion stimuli. Subsequent prediction analysis revealed ten features (i.e., cortical responses from eight frontocortical fNIRS channels, age, and sex) strongly correlated with PTSD severity. Examining if similar activation patterns and models predict treatment response would be an important next step in testing the utility of activation patterns to predict treatment response (Weems et al., 2019; Weems & Neill, 2020).

In this study, we conducted a randomized clinical trial comparing TF-CBT, CCT and Treatment as Usual (TAU) among youth in a community outpatient setting. We assessed participants at three time points: pre-intervention (T1; less than one week before intervention started), post-intervention (T2; less than one week after intervention concluded, and follow up (T3; three-months after conclusion of intervention). We hypothesized that 1)across conditions there would be treatment benefits that would vary by individual but that 2) CCT and TF-CBT would produce relatively better treatment outcomes than TAU, and 3) similar activation patterns found to predict PTSD symptom severity using fNIRS would also predict PTSD symptom treatment response.

2. Method

3. Results

4. Discussion

In this study we conducted a small randomized controlled trial to compare the effectiveness of three separate interventions on PTSD symptoms and examine the utility of fNIRS in predicting improvement among youth. We hypothesized that across conditions there would be treatment benefits that would vary by individual but that 1) TF-CBT and CCT would produce relatively better treatment outcomes than TAU. Consistent with hypothesis 1, youths who received CCT and TFCBT reported significantly lower levels of both PTSD and anxiety by follow-up while those in the TAU group indicated no significant differences over time. Additionally, there was no reduction in depressive symptoms in any of the three groups. With respect to hypothesis 2, the results showed that task-evoked response from six fNIRS channels within the dlPFC allow prediction of PTSD symptom improvement. These findings add to the emerging fNIRS literature demonstrating the feasibility of assessing cortical signatures of symptomatic improvement in an affordable and portable manner, and specifically underscores the utility of this approach for youth with posttraumatic symptoms.

The PTSD symptom reduction seen among youth receiving TF-CBT is consistent with previous studies. In a meta analysis of 21 randomized controlled trials of TF-CBT versus a waitlist control group or active treatment, Lenz and Hollenbaugh (2015) demonstrated large to moderate effect sizes, respectively. Additional analyses indicated those who receive TF-CBT also reported decreases in depressive symptoms, although these effect sizes were in the moderate to small ranges in waitlist control and active comparison treatments, respectively. Comparisons to previous research on CCT are less available, however. In the only other trial conducted to date, Carrion et al. (2013) found CCT was more effective than a waitlist control. The findings from the present study provide further support for the effectiveness of CCT, but future trials are necessary to continue developing the evidence for this approach.

Although both TF-CBT and CCT were effective in reducing PTSD symptoms, those in the TAU condition also reported some level of symptom reduction. This finding is consistent with previous research conducted in community-based settings such as outpatient clinics. For example, in a meta-analysis of 52 randomized controlled trials, Weisz (2013) and colleagues found lower effect sizes for evidence-based youth psychotherapies compared with usual care. This effect also became non-significant when accounting for how participants were enrolled (clinically referred vs youth presenting to clinics with diagnoses). In a large randomized effectiveness trial of evidence-based approaches for youth anxiety, depression, trauma, and conduct problems conducted by community clinicians from multiple outpatient clinics, the evidence-based psychotherapy approach was not superior to usual care on any clinical measure (Weisz et al., 2020). Consistent with these findings, the meta -analysis conducted by Lenz and Hollenbaugh (2015), effect sizes for PTSD symptom improvement were also smaller when compared against active, alternative treatments. These findings do not suggest that TF-CBT or CCT are no more effective than TAU, but that the lack of robust differences in our results is consistent with other studies conducted in community-based settings.

Results further demonstrated that PTSD symptom improvement could be predicted via fNIRS cortical signals. Specifically, results showed that the emotion-evoked response from six fNIRS channels in the bilateral dlPFC correlated with PTSD symptom improvement (r = 0.53).

These findings align with fNIRS and fMRI studies that associate the bilateral dlPFC with pediatric and adult PTSD (Balters et al., 2019; Fonzo et al., 2017; Garrett et al., 2012) and demonstrate associations between treatment-related PTSD symptom reduction and normalization of dlPFC brain function in adults (Matsuo et al., 2009; Fonzo et al., 2017). With respect to prediction performance, there is, to our knowledge, only one other study that has established association between fNIRS cortical signatures and PTSD symptom improvement (Matsuo et al., 2009). The study focused on adults with PTSD and found a significant positive association between the longitudinal change in cortical activation (i.e., difference between pre- and post-treatment) in the lateral PFC and PTSD symptom improvement (r = 0.66, p = 0.02, N = 12). A likely reason for the reduced prediction performance in this study could be that the present data was collected at mid-treatment and therewith potentially decreased the effect of symptom improvement. An additional explanation could be that Matsuo et al. (2009) used recall events as stimuli which could potentially result in increased fear responses compared to our facial expression task (Miller et al., 2021). Lastly, Matsuo et al. (2009) used an adult sample, and our youth sample could contain more variance in cortical responses that are associated with brain development (Keding & Herringa, 2016; Herringa, 2017). Future research should consider these experimental variables and create additional benchmarks of PTSD cortical signatures in youth. Overall, our findings suggest that there are associations between fNIRS cortical signatures and PTSD symptom improvement in youth, and that task-related fNIRS assessments are a promising direction for the development of effective prediction models of PTSD symptom improvement.

While the predictive model derived the optimal feature set to best predict PTSD symptom severity on the given dataset, the model does not provide information about the underlying cortical activation pattern. The optimal feature set could therefore represent a combination of cortical activation pattern that could include both increased and decreased responses to the facial stimuli. To better understand the underlying cortical mechanisms of the prediction model results, the associations between cortical activation and PTSD symptom improvement for each of the six channels were inspected. Results showed overall positive correlations between the activation of the channels in left dlPFC (i.e., channels 2,9,10,11) during fearful and neutral faces and PTSD symptom slopes. In other words, higher symptom improvement (i.e., more negative PTSD symptom slopes) was associated with lower cortical responses to fearful and neutral faces. These findings align with previous studies that identified dysregulated brain functioning during negative emotion processing and regulation in the left dlPFC in association with adult PTSD (Fonzo et al., 2017). The dlPFC is part of a prefrontal circuitry that is known to be involved in emotion regulation (i.e., top-down control over amygdala; Ganella et al., 2017; Gee, 2016). Reduced activation of the left dlPFC in individuals with higher PTSD symptom improvement therefore suggest that these individuals had a more normalized threat response (i.e., diminished activation of the frontoamygdala circuity to facial stimuli). In this context it is important to note that there is increased evidence that youth with PTSD, in contrast to adults with PTSD, exhibit fear responses not only to fearful faces but also neutral faces (Garrett et al., 2012; Garrett et al., 2019; Keding and Herringa, 2016; Balters et al., 2021). Underlying reason for the fear response has been attributed to the ambiguity of neutral faces that could be potentially perceived as threatening to youth with PTSD (Garrett et al., 2019). Our findings regarding the left dlPFC, thus, indicate an association between normalization of brain function within this region and PTSD symptom improvement.

In contrast to the positive association between PTSD symptom slope and activation in the left dlPFC, findings are overall indicative of negative associations between the right dlPFC and PTSD symptom slopes in response to neutral faces. In other words, increased symptom improvement (i.e., more negative PTSD symptom slope) was associated with higher activation in the right dlPFC. The right dlPFC is known to be involved in cognitive attention control (Kondo et al., 2004; Sanchez-Lopez, 2018) and researchers have argued for the causal role of the right dlPFC in the generation of attentional impairments that are implicated in emotional disturbances such anxiety (Sanchez et al., 2016). Because “problems in concentration” is a common DMS-5 symptom (American Psychiatric Association, 2013) in youth with PTSD symptoms (Russel et al., 2017), our findings suggest that PTSD symptom improvement leads to normalization of cortical function in the right dlPFC. Individuals with higher levels of PTSD symptom improvement utilized more cognitive resources in the right dlPFC that potentially afforded them with higher attentional control (i.e., higher levels of concentration). In the context of pediatric PTSD symptomatology our results show one more interesting finding. In our previous study that utilized fNIRS to predict PTSD symptom severity, the right middle temporal gyrus (MTG) was identified as an important prediction feature (Balters et al., 2021). In our present study, the MTG was not detected as an important prediction feature. Activation in the right MTG has been associated with dissociation, another key symptom in PTSD (Lanius et al., 2002; 2005). The absence of the MTG as important prediction feature of PTSD symptom improvement is a further sign of normalization of aberrant cortical brain function in pediatric PTSD. Lastly, in contrast to the previous findings that also identified age and sex as important prediction features of PTSD symptom severity, the present findings did not determine age and sex as an important prediction feature. While the underlying cause for the absence of age and sex as prediction features could be the reduced sample size in this study (n = 31 vs. n = 56), the results could also suggest that PTSD-related alternations in brain functioning are sex- and age dependent, and normalize with symptom improvement. Future research needs to be conducted to systematically study the effect of age and gender in the context of pediatric PTSD symptom improvement.

Although these data provide a foray into the utility of fNIRS to measure and predict treatment outcomes among youth with history of TRACEs and PTSD symptoms, there are several important limitations. The small sample size limited statistical power in our larger predictive model and treatment outcome comparisons among groups. Premature termination in youth mental health care ranges from 28%-75% and is even higher in effectiveness than in efficacy studies (de Haan, Boon, de Jong, Hoeve, & Vermeiren, 2013). Although common in effectiveness trials conducting treatment research in community mental health settings (e.g., Chorpita et al., 2015; Weisz et al., 2020), loss of data due to drop out at post-intervention and follow-up lowered our statistical power and ability to draw conclusions about treatment outcomes. Future studies examining cortical activation using fNIRS among youth with TRACEs should incorporate larger samples to replicate these findings. Additionally, concomitant fNIRS and fMRI imaging could be conducted to corroborate fNIRS findings and further elucidate how cortical activity measured during fNIRS corresponds to activity in subcortical areas. Future work should also incorporate age and sex-matched healthy controls (i.e., youth without a history of TRACEs) to examine how patterns of activation measured using fNIRS may differ before and after treatment.

Despite these limitations, our study was the first to use fNIRS to examine changes in cortical activation among youth with history of TRACEs receiving evidence-based approaches to treat PTSD symptoms. Given the barriers to traditional (i.e., fMRI) imaging methods, fNIRS provides a cost-effective, flexible method for measuring cortical activity. We also conducted fNIRS imaging within a community clinic among typical treatment seeking families. This approach creates opportunities to collect functional imaging data among a more ecologically valid treatment seeking sample compared to traditional (e.g., university hospital, academic medical center) settings. Our sample was highly heterogeneous and diverse, which allowed for greater external validity in interpreting results. Lastly, the differential improvement within TF-CBT and CCT also provides further support for these interventions for youth with PTSD symptoms.