Multiple axonal lesions of varying severity are scattered throughout the white matter in the brains of DAI patients, and various cognitive sequelae are observed depending on the severity and location of such lesions. Since PM function depends on intact intra-cerebral networks in several cognitive domains, disruption of these connections results in the failure of PM. DTI is a suitable method for detecting such disconnections, and is therefore considered useful for identifying certain domains associated with PM. We believe it is possible to determine the brain region attributable to each cognitive function affected in these patients by comparing cognitive parameters and FA values, which can be used to evaluate the severity of axonal injury at voxel level. With this hypothesis, we sought to locate lesions attributable to PM failure after DAI. To the best of our knowledge, no previous study has mapped the lesions attributable to PM failure through voxel-based lesion-symptom analysis using DTI. In this study, three clusters significantly correlated with PM failure were found in the left parahippocampal gyrus, left inferior parietal lobe, and left anterior cingulate.
Of the clusters correlated with PM failure, that in the left parahippocampal gyrus is consistent with the findings of previous neuroimaging studies with loading of experimental PM tasks. In previous PET studies, activation of the left parahippocampal gyrus was observed during loading of experimental PM tasks, and activation of this region was thought to play a role in recognition of cues triggering the performance of intended actions [6, 7, 10, 25]. The second cluster correlated with PM failure was detected in the left inferior parietal lobe. Voxel-based regression analysis with Trail Making Test-B, which was conducted in the same DAI subjects, showed that the cluster correlated with Trail Making Test-B was in the left inferior parietal lobe, in a location close to the cluster correlated with PM (Additional file 1and 2). In functional neuroimaging studies, cortical activation in this area was observed on loading of the Wisconsin Card Sorting Test and Trail Making Test, suggesting that neuronal activity in this region reflects the cognitive process of set shifting [26, 27]. Set shifting is believed to participate in PM, e. g. in monitoring of the environment for cues to re-instantiate an intention while performing other tasks (i.e. the second phase of the PM process) or inhibiting other activities at the critical time for performance of the intended action (i.e. the fourth phase of the PM process). It also appeared that lesions in the left anterior cingulate worsen PM function. This finding is consistent with a previous lesion-symptom analysis study using computed tomography. In that study, lesions in the left anterior cingulate causing PM failure were associated with failure of recall of the intention and intended action . These findings suggested that PM failure after DAI might reflect operation of the supporting retrospective components of PM, as well as those related more specifically to maintenance of an intention.
Previous studies using functional neuroimaging and event-related potential recording have revealed a relationship between the right inferior parietal lobe and PM function [1, 7–10, 28]. In the present study, a correlation between FA value and PM score was found for the cluster in the right inferior parietal lobe. However, post-hoc testing revealed that lower FA values in this cluster could be assigned to the cluster beyond the Sylvian fissure. Further study using accurate voxel-based morphometry is needed to determine whether atrophy of this region contributes to PM failure in patients with DAI.
The localization of the clusters correlated with PM failure was strikingly similar to that of regions activated during experimental PM tasks in previous activation studies, except for anterior prefrontal cortex (aPFC), which has been suggested to be of central importance to PM by several researchers [6–10]. On post-hoc testing, the reduction of FA value in white matter close to aPFC was not significant compared with that in normal controls (data not shown). It is likely that an artifact originating from air in the frontal sinus masked the lesion in aPFC determined by FA value.
The present study is the first to evaluate the neuronal basis of PM using voxel-based regression analysis with FA maps in DAI subjects. An advantage of our methodology is that no experimental PM task was required; we could instead adopt the widely used clinical PM assessment battery. However, among the regions elicited by our method, we could not discriminate that attributable to core PM from that attributable to supportive function for PM.
Other limitations of our study include the following. First, we examined a relatively small number of subjects. Future studies with larger samples of patients are needed to augment the reliability of our findings. Second, DTI data were acquired with diffusion weighting encoded along only 6 independent orientations. To obtain more accurate tensor parameters, a larger number of gradient sampling orientations must be adopted in further studies. Third, the interval between injury and testing varied widely between 3 to 51 months. In a previous study, longitudinal decline of FA values was observed in brains subjected to trauma , suggesting that longitudinal changes in FA values may have affected our results. However, post-hoc examination revealed no significant decline in FA value with time from injury. Hence, FA values were not adjusted for time from injury. Despite these limitations, we believe that we were able to map the neuronal basis for PM performance in a daily life situation.