Overall, these results show correlation between cognitive impairments and WM tract pathology as measured with DTI in bvFTD and AD. Furthermore, even after GM volumes adjacent to the WM tracts were included in the analysis, specific tracts’ FA remained a significant predictor of some tests of cognitive deficits in bvFTD.
The left anterior Cg tract FA remained a significant predictor of a number of executive tasks in the bvFTD group even after addition of GM volumes. Patients with bvFTD have significant degeneration in anterior cingulate regions [37
]. There are distinct differences in cytoarchitectonics, connections and function of the anterior and posterior cingulate gyrus. Anterior motor-related cortex is strongly interconnected with amygdala, nucleus accumbens, medial dorsal thalamus, and dorsolateral prefrontal cortex, whereas the posterior, sensory-related cortex is interconnected with temporal association cortex, medial temporal cortex, and parietal and orbitofrontal cortex. The anterior division is believed to be integral to the processing of error detection and decision monitoring [17
Spatial Span, an executive task of visual-spatial attention and working-memory was significantly related to the right anterior Cg even after addition of the right anterior cingulate GM volume. This task has a significant working memory component and so relies on intact executive function. Visual-spatial processing is usually attributed to the right hemisphere [12
]. A PET study revealed the right dorsolateral prefrontal cortex is important for Spatial Span function [6
The various executive tasks in this study are not necessarily taxing the same neuroanatomical regions. The Trails Composite score of DKEFS Trails Shifting is the only task that requires coordination between set-shifting as well as visual search and motor performance [24
] and in one study was the only test that showed correlation between frontal FA and cognitive task in AD patients [39
]. The anterior cingulate is implicated in error detection and performance monitoring [10
]. Total Executive Function Errors was no longer significantly related to the left ant Cg when the left anterior cingulate GM volume was added to the model indicating that it did not make any independent contribution. However, MT time, a measure of reaction time on an executive task, was related to the left ant Cg FA even with addition of GM volume.
A number of the executive tasks in this study were not predicted by any of the WM tracts. The reason may be that the tasks do not make the same demands on frontal lobe function. Design Fluency has been correlated with more widespread gray matter atrophy [31
Interestingly, Block design, a task thought to test visualspatial function, was significantly predicted by the right posterior Cg tract FA even after addition of the posterior cingulate GM volume. This test relies heavily on parietal function and patients with bvFTD usually demonstrate relative preservation of visual-spatial function. Kaplan [20
] proposed four explanations for difficulty with task: (i) loss of overall configuration due to distortion of gestalt (right hemisphere) [20
]; (ii) incorrect block placement within a grossly correct gestalt (left hemisphere) [19
]; (iii) impulsive, disorganized approach (frontal lobe); (iv) perseverative responding and cognitive inflexibility (frontal lobe). It may be that executive dysfunction and difficulty with actual block placement may be responsible for the relationship observed between the right posterior Cg and this task while visual-spatial function and gestalt perception are relatively intact.
The Modified Rey drawing, a simplified relative to the standard Rey-Osterreith Figure, the VOSP and calculations were not related to any tract. Parietal lobe damage is implicated in deficits on the VOSP and so the superior longitudinal fasciculus (SLF) may play a more important role. Mental calculations in a PET study activated a widespread anatomical network, which includes aspects of attention, auditory, and motor processing and working memory [13
]. One could presume that tract injury involving the SLF might have more of an impact on calculation. The working memory component is eliminated in our study because they are written calculations.
The number of MT errors was most related to the left Unc, which is part of the salience network and postulated as important for attributing value to rewards. Injury to this tract may lead to disinterest or lack of motivation to correct mistakes. The Trails task is the only executive task with feedback and the inability to attribute value to performing well may be brought out by this task and thus explain its unique relationship to the left Unc FA.
Recall of the Modified Rey diagram tests visual memory and was related to the right descending Cg. These fibers that cross through the parahippocampal gyrus, could be implicated in recognition of scenes as this area is important for topographical learning [1
]. Visual processing of Rey figure has been ascribed to right hemisphere [7
Adding the GM volumes adjacent to the WM tracts reduced significance of Trails Composite score of DKEFS Trails Shifting, Total Executive Function Errors and Delayed recall in the bvFTD. One explanation is that GM contributes more to certain cognitive tasks. MT time, a measure of speed, might be more sensitive to WM injury than cortical damage while the inverse might be true for MT number of correct lines, a measure of accuracy, which may require a more significant cortical component. Delayed visual recall may rely more heavily on the hippocampus and parahippocampal gyrus [22
] than any WM tract.
In the AD group, visual-spatial recall was related to FA in bilateral posterior Cg tracts and in the right descending Cg. When the GM volumes were added to the regression, none of the tracts remained significant predictors of visualspatial recall. The AD group was small and so there was less power to detect a relationship between WM tracts’ FA and tasks of cognitive function.
None of the language measures evaluated related to the WM tracts studied. This is likely because the left SLF, the most important language tract, was not tracked in this study. Since these patients have bvFTD, and usually face dysexecutive deficits and less often language, that tract was not included in this study. Verbal memory was also not related to the tracts and could be due to the strong hippocampal influence on verbal memory.
In regards to structural differences between the two groups, only the anterior corpus callosum FA and right uncinate GM were significantly different with both these measures lower in the bvFTD group. There were, however, a number of areas that showed trends that may have been significant with larger numbers including: right descending Cg FA lower in AD, right Unc FA lower in FTD, left anterior Cg GM lower in bvFTD, right Post Cg lower in AD. These differences although not statistically significant likely contribute to the cognitive differences seen in these two groups. These results are similar to previous pathological and radiological findings of gray matter and WM pathology in bvFTD and AD [8
]. Focal degeneration in bvFTD has been demonstrated in the rostral and subgenual anterior cingulate cortex, frontoinsular region, and the frontal pole [35
]. In the case of AD, the posterior cingulate region and medial temporal regions are early targets in the disease [3
]. Classically, decrease in FA has been used as a marker of myelin injury with axonal loss. Alternatively, fiber reorganization, glial alterations, increased membrane permeability and diffusivity, destruction of intracellular structures, alterations in the cytoskeleton and axonal transport, could also occur in neurodegenerative disease, and could influence DTI [4
]. At this time it is unknown whether WM pathology in neurodegenerative disease is a primary process, independent of GM pathology, or there is secondary reduction of WM integrity because of wallerian degeneration. As we learn more about the pathology that underlies the abnormal FA signal seen with DTI, this measure becomes another potential biomarker for detecting neurodegenerative disease. With a greater understanding of DTI’s pathological correlation, one can envision using a tract’s FA to look for concomitant WM and GM abnormalities that could be contributing to the cognitive/behavioral deficits seen in these illnesses or to stage a disease if it turns out that WM pathology occurs before or after GM injury.
In summary, this study demonstrates preliminary evidence within a small sample that the reduced integrity of specific WM tracts contributes to cognitive deficits observed in bvFTD after accounting for GM atrophy. Further study is required to assess whether WM pathology is a primary process in the neurodegenerative process or secondary to GM pathology.