[ARTICLE] Neuroplasticity of cognitive control networks following cognitive training for chronic traumatic brain injury – Full Text

Abstract

Cognitive control is the ability to coordinate thoughts and actions to achieve goals. Cognitive control impairments are one of the most persistent and devastating sequalae of traumatic brain injuries (TBI). There have been efforts to improve cognitive control in individuals with post-acute TBI. Several studies have reported changes in neuropsychological measures suggesting the efficacy of cognitive training in improving cognitive control. Yet, the neural substrates of improved cognitive control after training remains poorly understood. In the current study, we identified neural plasticity induced by cognitive control training for TBI using resting-state functional connectivity (rsFC). Fifty-six individuals with chronic mild TBI (9 years post-injury on average) were randomized into either a strategy-based cognitive training group (N = 26) or a knowledge-based training group (active control condition; N = 30) for 8 weeks. We acquired a total of 109 resting-state functional magnetic resonance imaging from 45 individuals before training, immediately post-training, and 3 months post-training. Relative to the controls, the strategy-based cognitive training group showed monotonic increases in connectivity in two cognitive control networks (i.e., cingulo-opercular and fronto-parietal networks) across time points in multiple brain regions (p_voxel < 0.001, p_cluster < 0.05). Analyses of brain-behavior relationships revealed that fronto-parietal network connectivity over three time points within the strategy-based cognitive training group was positively associated with the trail making scores (p_voxel < 0.001, p_cluster < 0.05). These findings suggest that training-induced neuroplasticity continues through chronic phases of TBI and that rsFC can serve as a neuroimaging biomarker of evaluating the efficacy of cognitive training for TBI.

1. Introduction

A traumatic brain injury (TBI) occurs when external force is applied to the head leading to disruptions of brain structure and function (Faul et al., 2010). Though an insult to the brain occurs instantaneously, a TBI incident can be the beginning of a chronic disease process rather than an isolated event or final outcome across all levels of initial injury severity: moderate or severe TBI (Corrigan et al., 2014; Masel and DeWitt, 2010; Whitnall et al., 2006) and mild-to-severe TBI (Masel and DeWitt, 2010; Whitnall et al., 2006). For example, TBI can be a risk factor for cognitive impairments (Arciniegas et al., 2002; Rabinowitz and Levin, 2014), psychiatric disorders (Hesdorffer et al., 2009), reduced social functioning (Temkin et al., 2009), and neurodegenerative diseases such as chronic traumatic encephalopathy (McKee et al., 2013). A substantial number of individuals with TBI sustain TBI-related disabilities. For example, 57% of individuals 16 years or older with moderate or severe TBI were moderately or severely disabled, and 39% had a worse global outcome at 5 years post-injury compared to their outcome level at 1 or 2 years post-injury (Corrigan et al., 2014). Currently, as many as 5.3 million people in the U.S. are facing challenges of TBI-related disability (Frieden et al., 2015). The actual number of individuals continuing to suffer from chronic TBI (>6 months post-injury time) effects may be greater than the estimates given the lack of public awareness of TBI in the past and the limited sensitivity of conventional neuropsychological measures (Katz and Alexander, 1994). Additionally, conventional clinical imaging (e.g., CT scanning) may be insensitive to identifying brain abnormalities especially in individuals with mild TBI (Tellier et al., 2009). Substantial numbers of individuals with sustained TBI necessitates further rehabilitation research in chronic TBI (Katz and Alexander, 1994).

Resting-state functional connectivity (rsFC) is a technique measuring the temporal coherence of blood oxygenation level dependent (BOLD) signal from anatomically separated brain regions acquired at rest. Since its inception (Biswal et al., 1995), rsFC in resting-state functional magnetic imaging (rsfMRI) has provided new insights about brain networks that can better explain the underlying mechanisms of human behavior or function (van den Heuvel and Hulshoff Pol, 2010). RsFC studies in clinical populations are increasingly popular because they do not require that subjects perform a specific task. RsFC is well-positioned to identify both the patterns of injury and the associations between injury and behavioral impairments in TBI (Sharp et al., 2014). This is especially important as diffuse axonal injury (DAI) is one of the primary injury mechanisms of TBI (Smith et al., 2003). DAI induces multi-focal injuries to axons which provide the structural basis of spatially distributed brain networks. Thus, DAI leads to a breakdown of brain network connectivity. In the context of rehabilitation, rsFC is also a promising technique to measure neuroplasticity within the injured brain, as rsFC has been successfully utilized to provide evidence for experience-induced neuroplasticity of the adult human brain in vivo ( Guerra-Carrillo et al., 2014; Kelly and Castellanos, 2014). For example, in healthy subjects, previous studies reported changes in rsFC after motor training (Lewis et al., 2009; Taubert et al., 2011), cognitive training (Jolles et al., 2013; Mackey et al., 2013; Takeuchi et al., 2013), and physical activity in older adults (Voss et al., 2010). In clinical populations, changes in rsFC after cognitive rehabilitation for cognitive symptoms associated with multiple sclerosis has been reported (de Giglio et al., 2016; Keshavan et al., 2017). This technique is well-suited to investigating neuroplasticity induced by rehabilitation for TBI.

In a previous study, we reported the efficacy of strategy-based cognitive training for chronic TBI, utilizing neuropsychological measures (Vas et al., 2016). This training is an integrative program to improve cognitive control by exerting more efficient thinking strategies for selective attention and abstract reasoning (see the Materials and methods section for the details of training protocols). Cognitive control (also called executive function) is the ability to coordinate thoughts and actions to achieve goals while adjusting these goals according to changing environments (Nomura et al., 2010). Cognitive control is critical to successfully perform daily life tasks (Botvinick et al., 2001; Diamond, 2013). Thus, impairment in cognitive control is one of the most persistent and devastating sequalae of TBI (Cicerone et al., 2000; Rabinowitz and Levin, 2014), and empirical studies demonstrating the efficacy of cognitive rehabilitation for improving cognitive control of individuals with post-acute TBI are valuable in the literature on TBI rehabilitation (Cicerone et al., 2006; McDonald et al., 2002). In the current study, we describe rehabilitation-induced changes in brain connectivity.

Cognitive control has been extensively investigated in the field of cognitive neuroscience (Power and Petersen, 2013). Of note, Dosenbach and colleagues (Dosenbach et al., 2006) identified a set of regions that are active across multiple cognitive control tasks. A follow-up study (Dosenbach et al., 2007) revealed two distinct resting-state networks related to cognitive control: the cingulo-opercular network and fronto-parietal network. The cingulo-opercular network consists of bilateral anterior insula/frontal opercula (aI/fO), bilateral anterior prefrontal cortices (aPFC), dorsal anterior cingulate cortex (dACC), and thalamus, and it is thought to support stable maintenance of task mode and strategy during cognitive processes (Dosenbach et al., 2007 ;  Dosenbach et al., 2008). The fronto-parietal network comprises of bilateral dorsolateral prefrontal cortices (dlPFC), bilateral dorsal frontal cortices (dFC), bilateral inferior parietal lobules (IPL), bilateral intraparietal sulci (IPS), middle cingulate cortex (mCC), and bilateral precunei (PCUN), supporting active, adaptive online control during cognitive control processes (Dosenbach et al., 2007 ;  Dosenbach et al., 2008). The cingulo-opercular network and fronto-parietal network are also referred to as the salience network and central executive network, respectively (Seeley et al., 2007). The salience and central executive networks are often referred to in the context of interactions among these networks and the default mode network (Menon and Uddin, 2010). However, in this report, we will refer to them as the cingulo-opercular and fronto-parietal networks, as we conducted current study in the context of cognitive control. TBI-induced disruptions to the cingulo-opercular in mild-to-severe TBI (Bonnelle et al., 2012; Jilka et al., 2014; Stevens et al., 2012) and fronto-parietal networks in mild TBI (Mayer et al., 2011; Stevens et al., 2012) have been previously reported. Specifically, TBI decreases the white matter integrity of the cingulo-opercular network (Bonnelle et al., 2012) and functional connectivity between the cingulo-opercular and default networks during a cognitive control task (Jilka et al., 2014). Additionally, individuals with mild TBI showed increases and decreases in rsFC with the cingulo-opercular (Stevens et al., 2012) and fronto-parietal networks (Mayer et al., 2011; Stevens et al., 2012) across brain regions, relative to healthy individuals.

We utilized rsfMRI to identify the effects of a strategy-based cognitive training for chronic TBI on the cognitive control networks (i.e., cingulo-opercular and fronto-parietal networks) compared to a knowledge-based comparison condition. We focused on the cingulo-opercular and fronto-parietal networks as our training protocols were aimed at improving cognitive control processes (See the Materials and methods section for the details of training protocols). We randomized individuals with chronic mild TBI into two eight-week training groups (strategy- versus knowledge-based), and we acquired their MRI scans over three time points (prior to training, after training, and at three-months follow-up after training completed). We then investigated the spatial and temporal patterns of training-induced changes in cingulo-opercular and fronto-parietal networks connectivity of these individuals. We hypothesized that strategy-based cognitive training would induce changes in the cingulo-opercular and fronto-parietal networks connectivity relative to the knowledge-based training program. This prediction is based on findings from previous rsfMRI studies demonstrating neuroplasticity in healthy adults and other clinical populations (de Giglio et al., 2016; Jolles et al., 2013; Keshavan et al., 2017; Lewis et al., 2009; Mackey et al., 2013; Takeuchi et al., 2013; Taubert et al., 2011; Voss et al., 2010) and the efficacy of strategy-based cognitive training for chronic TBI (Vas et al., 2016).[…]

 

Continue —> Neuroplasticity of cognitive control networks following cognitive training for chronic traumatic brain injury

 

Fig. 1. Seed locations. Black and yellow circles represent seeds for the cingulo-opercular network and fronto-parietal network, respectively. aI/fO, anterior insula/frontal operculum; aPFC, anterior prefrontal cortex; dACC, dorsal anterior cingulate cortex; dFC, dorsal frontal cortex; dlPFC, dorsolateral prefrontal cortex; IPS; intraparietal sulcus; mCC, middle cingulate cortex; PCUN, precuneus; L, left; R, right.