Variation in the Developing Brain and the Role of Pediatric Posttraumatic Stress on Structural and Functional Networks

Introduction

Research using MRI imaging techniques has elucidated many neurobiological correlates of adverse and traumatic childhood experiences in youth. The Kribakaran and colleague meta-analysis (1) showed that pediatric PTSD is associated with relatively smaller volumes in total cerebral volume and gray matter as well as in number of regions including temporal lobe (total, right, left), total cerebellar vermis, and hippocampal (total, right, left) volumes than those without PTSD. Their meta-analysis also shows that age is significant moderator of variability for some of these regions such as hippocampal volumes and also amygdala regions when tracing studies were tested separately. Our perspective is that the findings indicate the need to employ a neurodevelopmental network approach to understand the relationship between PTSD and brain development. This commentary thus builds upon the meta-analytic findings (1) to discuss the issues and implications involved in taking a truly developmental and holistic approach to understanding the structural and functional neurobiological effects of trauma on children and youth.

An important starting point is the finding that total volume size is consistently associated with pediatric traumatic stress and PTSD. While we do not know how much of this correlation stems from preexisting differences - driven by poverty or other socioeconomic context effects that place youth and greater risk for experiencing trauma or developing interfering PTSD symptoms after traumatic stress - we start from the theory that the effect is indeed trauma related. How then should we understand the whole brain or total gray matter differences in terms of individual brain regions? Regional effects are typically tested (and found) when controlling for whole brain or total gray volumes (2). We do not empirically know about the other direction – if whole brain/gray differences remain controlling for sub-regions with prominent reductions in volume among youth. That is, are differences in total brain/gray volumes purely driven by reductions in key areas such as the temporal lobe, hippocampus, amygdala, etc., or are these changes truly ‘brain-wide’? The predominant theory of how stress impacts the brain is the hypothesis that severe or traumatic stress may damage particular regions of the brain. The initial work targeted the hippocampus, where early research suggested that glucocorticoids secreted during stress may damage pyramidal cells or suppress neurogenesis of granule cells (3).

Early theory and research on the role of traumatic stress on the brain began before the complex picture of regional brain development was fully appreciated. Recent work in large normative samples highlights this point (4). With findings integrating longitudinal design and testing effects across pubertal development indicating that the amygdala and hippocampus are generally larger at later Tanner stages with a tapering off of growth in stages four and five. However, the caudate, globus pallidus, nucleus, accumbens, and putamen decrease in size (4). Moreover, there is evidence that the connections between, for example, the amygdala and prefrontal cortex continue developing through adolescence and early adulthood, and that the connectivity of the amygdala becomes stronger and more differentiated from late childhood to early adulthood (5). While the knowledge base on normative trajectories of regional brain development has grown, a truly developmental whole brain network approach to the neurobiological effects of traumatic stress has not been the norm (6). Such an approach recognizes the parts in the context of the larger system by acknowledging the myriad links between global and regional brain changes, and specific symptoms in the context of the trauma or diagnosis (7).

Weems (6) developed a theory that traumatic stress exerts its effects on the brain as a moderator of normal developmental trends. This theory holds that age/maturation as the critical driver of variation in brain and severe stress influences the role or normal maturation. Moreover, while PTSD is clearly linked to particular regions (e.g., the amygdala, hippocampus) (1, 2), the brain is a complex networked system of many regions with essential structural and functional roles in optimal role in behavior, emotion, cognition, etc. The association between traumatic stress or PTSD diagnosis and differences in the size of various regions should be understood in terms of how these differences relate to the development of differential structural and functional networks in the brain writ large (7). Drawing from the adult MRI literature, Akiki et al. (8) argued that PTSD can be characterized by a weak and hypoactive default mode network (difficulty calming/resting) and an executive network (ineffective control) that is weakened by a strongly connected and highly overactive salience network (low threshold for perceived saliency of stimuli).

In the he MRI literature, it is the norm to control for age by matching traumatized groups on age or controlling for age in the analyses as a covariate (7). To date, research examining the effects of development (age or index of pubertal maturation, such as Tanner stage) and neural connectivity is quite uncommon. In a recent review and literature search we (7) found that no structural studies and only three functional studies tested for a moderating effect of age or maturation. Within the three functional studies that did analyze and report age or maturation effects differential patterns emerged in all three (7). Thus, Similar to the view that severe stress influences the normal developmental trajectories of particular regions (6), research on pediatric samples suggests a similar effect on functional connections in youth may exist (7). That is, normal maturation is likely also a critical driver of variation in structural and functional connections with stress, thereby altering normal developmental trends, particularly in emotion processing regions.

To date no longitudinal studies have examined the differential effect of traumatic stress on diverse brain region development among youth at different stages of puberty. However, (see 7) data from a small sample of 15 youth exposed to traumatic stress and followed over a one year period suggests a stage by time interaction in the development of amygdala volumes. Specifically, less developed youth (earlier Tanner) showed increases in amygdala volumes across time, whereas those in later tanner stages exhibited decreases over time. Clarifying traumatic stress effect on brain development requires longitudinal investigation with indices of biological maturation. We suggest cohort-sequential designs (9) which test developmental change using ‘cohorts’ of youth across a distribution of age or pubertal development at study entry and longitudinal assessment to test actual change. Theoretically, age or level of pubertal development at first assessment, age/maturation at time of trauma exposure; repeated assessment for actual change as well as assessment of exposure to traumatic stress/trauma type and PTSD symptoms are all important design elements in future research (7).

Figure 1 summarizes several points from this commentary and is based on other reviews of the literature (2, 6, 7) and illustrates a developmental neurobiological network perspective. The brain image highlights regions linked to the response to severe stress. From a network perspective these represent likely “central” regions that may hold critical influence over stress response networks. These include the pre-frontal cortex (the illustration is limited to the orbitofrontal region to allow visibility of the anterior cingulate), amygdala, and hippocampus, but also the insula and anterior cingulate (the literature does link severe stress to other regions not depicted, see 2). Brain structure, function, and connectivity are all nested within normal development, as indicated by the large right pointing arrow in which the brain is situated.

Figure 1.

Illustrates several major points in this article. Five regions important in the response to severe stress, including the pre-frontal cortex (just the orbitofrontal shown so the cingulate is visible) amygdala, hippocampus, insula and anterior cingulate are shown. Brain structure and function (and their connectivity) is nested within normal development (large right pointing arrow) and traumatic stress may influence the brain directly (e.g., inflammation) or by effecting normal development (e.g., epigenetic influence). The structural and or functional connectivity (represented by the oval arrows connecting across these regions) has been linked (indicated by the arrows) to symptom severity and clusters of symptoms such as hyperarousal (the triangular symptom cluster). Employing a network symptom approach may help understand differential associations amongst symptoms (arrows pointing to connections between symptoms) and the centrality of certain symptoms (arrows pointing to symptoms) in pediatric PTSD. Whole brain differences in pediatric PTSD may be a function of theoretically or clinically important differences in structural and functional networks. Network symptom associations and model adapted from Russell et al. (10) and Weems et al. (7).

The Figure highlights that traumatic stress may influence the brain directly (e.g., inflammation, glucocorticoid based damage) or by influencing normal development (e.g., epigenetic influence) indicated by the arrows from traumatic stress to both the brain and the arrow representing brain development. Connectivity is represented by the cyclic arrows depicting links among brain regions, has been associated with PTSD symptom severity (7) and clusters of symptoms such as hyperarousal (shown by the arrow leading to the highlighted symptom cluster). In the lower part of the figure is depicted the network symptom associations adapted from Russell et al. (10). Symptoms of PTSD are represented as shapes/nodes and connecting lines/edges indicate associations between symptoms with line thickness reflecting the magnitude of association. Conceptualizing the findings of the Kribakaran and colleagues meta-analysis (1) within a network framework may enhance understanding of the differential associations amongst symptoms (arrows pointing to connections between symptoms), the centrality of certain symptoms (arrows pointing to symptoms) and how brain networks effect and are affected by traumatic stress, PTSD or PTSD symptom developmental cascades. For example, if traumatic stress alters one or more of the broad structural networks – whole brain differences may be from a theoretically or clinically important effect on one or more networks that appear to be whole brain differences.

In conclusion, whole brain, substructure, function, and connectivity are all nested within normal development. A fully developmental approach to future research considers on severe or stress as moderators of normal maturation in brain growth, connectivity, and their link to symptomatology (not just diagnosis of PTSD). It is likely that cross sectional design will continue and so it’s imperative to test age as a moderator. However, clarifying the nature of the effect of stress will require longitudinal and even experimental designs involving youth at diverse stages of development (7).