Brain damage after ischemic stroke can cause a greater variety of functional deficits. These deficits can be caused by direct damage to the cortices or fiber tracts, For example, injuries to the right inferior parietal lobe, superior temporal gyrus and inferior frontal gyrus (IFG) are frequently associated with neglect (Corbetta and Shulman, 2011), and lesions to the corticospinal tract (CST) can cause hemiplegia (Lo et al., 2010). In addition to the direct damage caused by lesions, indirect atrophy of lesion-related remote cortices has also been reported (Rowan et al., 2007; Gauthier et al., 2012; Fan et al., 2013; Zhang et al., 2014; Cheng et al., 2015). For example, a recent research has demonstrated that cortical atrophy in remote cortices is also correlated with the magnitude of residual motor deficits in chronic sub-cortical stroke patients (Gauthier et al., 2012). In addition to the evidence of secondary cortical atrophy, many early studies also reported secondary degeneration of remote white matter tracts after damage to the motor pathway due to sub-cortical stroke (Thomalla et al., 2004, 2005; Liang et al., 2008; Yu et al., 2009; Rüber et al., 2012), and the severity of the degeneration predicts poor motor recovery (Yu et al., 2009; Lindenberg et al., 2010). These findings indicate that the secondary neurodegeneration of the motor pathways might be responsible for the atrophy of remote cortical regions and might consequently influence motor performance.

Although secondary structural impairment of remote cortex has frequently been reported, the structural plasticity of the remaining cortex after stroke has yet to be fully clarified. In a review of previous neuroimaging literature that focused on the cortical changes after stroke, we found that the majority of studies adopted either a retrospective or cross-sectional design, and heterogeneities in lesion location and duration were frequently present (Schormann and Kraemer, 2003; Kraemer et al., 2004; Schaechter et al., 2006; Rowan et al., 2007; Stebbins et al., 2008; Gauthier et al., 2012). Individual variability in cross-sectional studies might mask subtle changes in the cortex or induce some false positive results, and lesion heterogeneity increases the complexity of interpreting the underlying neuronal mechanism. In a recent study of longitudinal changes in cortical thickness 3 months after sub-cortical stroke (Brodtmann et al., 2012), the authors found thickening of the contralesional (CL) cortices; however, these authors did not find any atrophy of the ipsilesional (IL) cortices. In another similar longitudinal study, Cheng et al. (2015) failed to identify any changes in cortical thickness in the CL lesion-connected or lesion-unconnected cortices, although they observed a strong decrease in the cortical thickness of the IL lesion-connected cortex. The discrepancy between the two studies may have been caused by the following factors: (1) the relative shorter follow-up duration (3 months after stroke); (2) the insensitivity of the region-of-interest (ROI) analysis method; and (3) the constraint of the cortical thickness in completely characterizing cortical atrophy and plasticity without accounting for changes in cortical surface area.

In this study, we recruited a subset of motor-deficit stroke patients with first onset, subcortical ischemic infractions that involved the basal ganglia regions. In contrast to early longitudinal studies (Brodtmann et al., 2012; Cheng et al., 2015), we used a whole-brain voxel-based morphometry (VBM) method to identify the potential changes in gray matter volume (GMV) after stroke. Moreover, the follow-up duration in the present study was extended to 1 year. Because the GMV contains information about both cortical thickness and cortical surface area, any changes in these two metrics can be reflected by the GMV. Based on the features of the VBM method and the longer follow-up duration, we hoped to identify both atrophy and augment in the GMV of the remote cortices after 1 years. Specifically, based on early studies that revealed secondary cortical atrophy of the IL cortices and its association with clinical performance, we hypothesized that the GMVs of the IL motor-related cortices would be decreased after 1 year, and the severity of the atrophy of these cortices would be associated with clinical recovery. Because early studies also demonstrated wide-spread of functional reorganization of multiple brain network (Wang et al., 2010, 2014; Rehme et al., 2012), we also hoped to observed increases in the GMVs of the remaining cortices and significant association between GMV increases and clinical recovery.

In accordance with recent VBM studies (Yabuta and Callaway, 1998; Gauthier et al., 2012), we found GMV decreases after stroke in some intact remote cortices that were directly or indirectly connected with the basal ganglia regions through cortical projection fibers (e.g., the CST, cortical-striatal-thalamic pathway, and cerebro-ponto-cerebellar circuit). Focal damage in these pathways can lead to dysfunction in neuroanatomically related remote brain regions soon after ischemia. Region specific diaschisis in the CL cerebellum after stroke has been identified (Lin et al., 2009), and might be correlated with clinical outcomes (Szilágyi et al., 2012). Similarly, secondary neurodegeneration has been identified in earlier studies using diffusion tensor imaging (DTI) in sub-cortical stroke patients (Liang et al., 2008; Crofts et al., 2011). Finally, diaschisis and degeneration can induce secondary atrophy in remote cortices (Yabuta and Callaway, 1998; Cheng et al., 2015) as observed in present study. These findings indicate that neurodegenerative atrophy in these “healthy” areas might partially account for permanent motor deficits. Except for the CBVII, we didn’t observe any significant atrophy in other “cognitive” regions that also have projections through the basal ganglia regions, and similar results have been reported in previous studies (Yabuta and Callaway, 1998; Rowan et al., 2007; Gauthier et al., 2012). One possible reason for this findings is that the motor function was predominantly involved in our patients, and thus the most significant atrophies occurred in motor-related areas.

So far, only a few studies identified GMV increases after stroke. Efficacious constraint-induced movement therapy has been found to induce increases in GMV in the bilateral sensorimotor cortices and hippocampi of chronic stroke patients (Woodhead et al., 2011). Fan et al. (2013) found increases in GMV ipsilesionally in the medial cingulate cortex (MCC), lingual gyrus and cerebellar lobe, contralesionally in the hippocampus and bilaterally in the precuneus via the monitoring the dynamic structural evolution after sub-cortical stroke (Yabuta and Callaway, 1998). Moreover, Brodtmann et al. (2012) observed thickening of the CL cortices 3 month after sub-cortical stroke. Animal experiments have demonstrated that the GMV increases in spared cortices may result from rehabilitation-induced increases in dendritic arborization and synaptic density (Holtmaat and Svoboda, 2009). In present study, we identified GMV increase in CL hemishpere, which was contradicted to a recent study that failed to identify any structural plasticity after sub-cortical stroke (Cheng et al., 2015). The positive findings of CL cortical structural plasticity in present study ascribe from: (1) longer follow-up interval: (1 year in present study vs. 3 month by Cheng et al., 2015); (2) different structural measurements (GMV in present study vs. cortical thickness by Cheng et al., 2015); and (3) whole brain voxel-wise search strategies. It should be noted that our findings with increased GMV were mainly located in CL prefrontal “cognitive” regions (MFG, OFC and IFG) rather than motor-related cortices (such as the premotor cortex and supplementary area; Yabuta and Callaway, 1998; Schaechter et al., 2006; Woodhead et al., 2011). In healthy subjects, the dorsal lateral prefrontal cortex is closely associated with attention (Corbetta et al., 2008), execution control (Goldman-Rakic, 1987) and motor learning (Tomassini et al., 2011), and the OFC is involved in motivational, emotional and social behavior (Rolls, 2004) and the integration for executive motor control (Cavada et al., 2000). Although the exact mechanism is far from clear, prior studies have demonstrated that cognitive abilities are particularly important for the recovery of motor function (Leung et al., 2010), and cognitive strategy-based interventions have beneficial effects on the recovery of motor function in stroke patients (Cirstea et al., 2006). Furthermore, a recent study demonstrated that a motor learning task can invoke greater activity in the dorsolateral prefrontal cortices in stroke patients with hemiplegia (Meehan et al., 2011). In this study, we also found that greater increases in the GMVs of these “cognitive” regions predicted better motor performances, indicating that the structural reorganization of these “cognitive” cortices may play an important role in motor recovery after stroke.

Several limitations should be noted. First, the sample size is relatively small, and the statistical power is thus low. To increase the statistical power, we right-left flipped the structure data of patients with right infarct lesions. Although this method was frequently applied in early stroke studies (Gauthier et al., 2008; Sharma et al., 2009; Li et al., 2013), it may be inappropriate for evaluating the brain areas with functional lateralization, such as the language and spatial processing areas. Because the number of patients with left-hemisphere lesions (seven cases) was much greater than that of the patients with right-side lesions (four cases), our findings (especially the increased GMV in the CL prefrontal cortex) may have primarily been contributed by patients with left-side infarctions. To clarify whether the patients with right-side infractions would exhibit patterns of change similar to those of the patients with left-side infarctions, we plotted the GMV changes subject-by-subject (Figure 3). We found that among the four right-side infraction patients (patient IDs: str01, str07, str08, and str11), with the exception of case str11, who manifested relatively weak GMV changes (but without the reverse changes), the remaining three patients exhibited change amplitudes comparable to those of the cases with left lesions. Thus, the laterality of the GMV changes may not have been significant in the present study. Of course, this assumption should be verified in the future with larger sample sizes. Second, we did not evaluate the cognitive functions at any stage in these patients, thus, the possibility that the augmentations in the GMVs reflected the compensatory plasticities of specific cognitive functions cannot be excluded.

In conclusion, by applying a longitudinal and lesion-restricted approach, we observed secondary neurodegeneration in remote cortices after sub-cortical stroke. We have also provided evidence of cortical plasticity in healthy cognitive-related cortices that might contribute to motor recovery. Further studies of adaptive plasticity in these remote areas may be important for developing and evaluating strategies for therapeutic intervention.

KL, JC and RP designed the experiment. JC, XN, QJ, DZ and RX performed the experiments. JC and QJ analyzed the data. JC, KL and RP drafted and revised the manuscript. All authors discussed the results.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.