The present study characterized, in vivo
, white matter changes in the parahippocampalregion that includes the perforant path in individuals with aMCI compared to those of elderly controls with no cognitive impairment using volumetry, DTI and tractography. The results from this study confirm and extend previous findings in our laboratory [50
] and others [17
], showing that compared to controls, individuals with aMCI have volume loss in the parahippocampal white matter. We also found a concomitant increase in MD in the aMCI group, for both the PWM region including the perforant path, as well as in the fibers modeled to pass through this PWM region. In addition to demonstrating group differences, regression models were used to examine the relationships between PWM changes and those of neighboring neuroanatomically connected gray matter structures (entorhinal cortex and hippocampus), as well as memory performance.
Total normalized EC, but not HF volume, was a significant predictor of total normalized PWM volume, suggesting that volume loss in the PWM may be related to cell loss in the EC. This finding has been reported by our laboratory [50
] and others [43
] and is consistent with the hypothesis that memory deficits that characterize individuals with aMCI and AD may partly be a consequence of a disconnection in the flow of information from the entorhinal cortex to the hippocampus via the perforant path [23
The relationship between FA and MD in the PWM region was explored to better understand microstructural changes in white matter measured by DTI. Typically, there is an inverse relationship between FA and MD, so that as tissue is damaged or atrophied, there is an increase in MD due to increased free diffusion, but there is usually a concomitant decrease in FA due to loss of directional diffusion because of a loss of barriers to isotropic diffusion. In the region of the perforant path, the aMCI group in this study did not show the expected relationship between MD and FA values. Their MD was increased compared to the NCI group, but there was no difference in FA between the groups. This finding suggests that although individuals with aMCI have general diffusivity alterations in the PWM region, the longitudinal microstructural organization of the remaining parallel fibers of the perforant path region, as measures by FA, is comparable to that of controls. This result is in contrast to a report of decreased FA in the PWM containing the perforant path in patients with AD [43
]. In that study, specific alterations in anisotropic diffusion parameters associated with myelin disruption [45
] were found. This disruption of myelin function in AD may be related to increased susceptibility of PWM oligodendrocytes to metabolic and free radical formation [4
The lack of inverse correlation between FA and MD, while less common, has been reported in individuals with MCI [17
], as well as in studies of patients with HIV [48
] and children with 22q deletion syndrome [44
]. Unfortunately, the mechanism for tissue disruption cannot be definitively determined using DTI. The neurodegenerative changes taking place in this region are exceedingly complex and the changes in MD may reflect a combination of processes including loss of axons, dendrites and tissue degradation due to anterograde Wallerian degeneration. This may be a consequence of the cell loss in layer II of the EC and is consistent with the EC volume loss reported here and by others [43
]. Controlled pathological studies will be important for disentangling the temporal course and mechanisms of the neurodegeneration.
A second aim of the study was to examine the contribution of the in vivo imaging measures from the PWM ROI to memory performance. Results from the ROI analyses revealed that MD in the white matter region containing the perforant path was the only significant predictor of memory performance, demonstrating the functional relevance of the diffusivity changes. Our results also suggest that the memory deficits that characterize individuals with aMCI do not require PWM region changes in FA. It is possible that FA may play a greater role in predicting memory performance as the disease progresses to meet the criteria for clinical AD. Future cross-sectional and longitudinal studies, comparing the PWM region integrity between the three diagnostic groups (NCI, aMCI and AD), are required to characterize the temporal changes in the perforant path region.
The tractography results indicated that both FA and MD of the fiber tracts generated from the PWM ROI contributed significantly to memory performance in the regression models. The significant addition fiber FA to the prediction of memory performance suggests the relationship between the fibers in the local PWM ROI and the fibers that pass through the PWM region is different. One caveat of tractography estimates from a single ROI seed point is that it only provides information about the fibers passing through this region and is not limited to fibers originating in the PWM region that includes the perforant path. Despite this limitation, it is important to note that the diffusion data of the tracked fibers corroborate the PWM ROI data, which show a consistent increase in MD, but no change in FA.
Though several studies have examined white matter changes in MCI and AD, few have targeted the perforant path as a region of interest using DTI [27
]. The results from these studies have been mixed. The inconsistent findings may be attributed to the methodological variability between studies, including different approaches to the acquisition and analysis of DTI parameters (e.g., diffusion weights, eddy current correction), inconsistency in the DTI metrics reported (e.g., coherence index, FA, MD, axial and radial diffusivity measures) [17
], and variability in the region of interest definitions (e.g. spherical, rectangular) [17
The diversity of methodologies used in previous studies limits generalizability and makes it difficult to compare the previous results to the present findings. For example, Kalus and colleagues [27
] manually traced the perforant path ROI but only examined the coherence index measure of DTI, leaving out more common measures of FA or MD from the analysis, thus preventing parallel comparison to the findings in the present study. The temporal lobe white matter DTI results from Fellgiebel and colleagues [17
] are consistent with those reported in this study, i.e., increased MD but no change in FA. However, their method of ROI placement (rectangular) was not as anatomically specific as our targeted manually traced ROI. The present study applied comprehensive eddy current corrections to the DTI images, used a clearly defined, anatomically driven protocol for segmenting the PWM region and reports data using the two most common DTI metrics, MD and FA. These methodological considerations ensured accurate representation of the white matter region and will allow for replication of the study by future investigators.
In summary, the present study used in vivo imaging to demonstrate regional tissue loss and degradation as evidenced by decreased volume and increased MD within the PWM region that includes the perforant path in individuals with aMCI compared to cognitively intact elderly controls. However, the parallel microstructural organization (as measure by FA) of the remaining fibers within the PWM region was similar between diagnostic groups. Further, it is not surprising that the DTI metric of MD, but not FA, was predictive of memory performance since FA was not significantly different between the aMCI and control groups. Taken together, these findings suggest that even with intact microstructural organization, tissue loss and degradation of the perforant path fibers contribute to the memory loss that characterizes individuals with aMCI who are at risk for developing AD. Though the exact mechanism cannot be determined, our data are in line with the hypothesis that parahippocampal white matter degradation causes a disruption of multimodal input from the entorhinal cortex to the hippocampus. It is important that future research characterizes the temporal changes in the microstructural organization of the PWM throughout the neurodegenerative process from aMCI to clinical AD.