Using TBM, we mapped and compared regional differences in the progression of brain atrophy over a 1-year period among patients with three clinical syndromes of FTLD. In comparison to the language variants, bvFTD subjects demonstrated significantly higher rates of frontal atrophy, though there was variability in the specific regions of difference. The bvFTD group exhibited greater atrophy than the NF-PPA group in the VMPFC, but showed greater atrophy in the inferior, middle and superior frontal gyri compared to SV-PPA. The SV-PPA group showed higher annual atrophy rates than bvFTD in the left inferior temporal lobes, including the fusiform gyrus, which corresponds to the language disturbances and prosopagnosia that are specific to SV-PPA. Greater expansion of the posterior horns of the lateral ventricles was also observed in SV-PPA compared to bvFTD. The patterns of regional atrophy reported in our SV-PPA group are very consistent with the literature, particularly the involvement of the fusiform gyrus [13
]. Other longitudinal imaging studies using TBM or similar approaches also described gray matter contraction in the temporal pole and increased rates of volume loss in the right inferolateral temporal gyrus of SV-PPA subjects compared to healthy controls, with sparing of parietal, occipital, and dorsal frontal brain regions [15
While the above results replicate previous reports on the distribution of brain atrophy in clinical subtypes of FTLD, a novel finding of our study is that the bvFTD subjects demonstrated significantly more longitudinal white matter contraction than the SV-PPA group in the superior and inferior frontal lobes. Even though select cross-sectional studies have shown white matter atrophy in FTLD on structural MRI [10
], the data on longitudinal white matter changes in FTLD is much more limited. White matter damage, specifically moderate frontal white matter gliosis with demyelination, has been documented pathologically in frontotemporal dementia patients [17
]. Broe et al. [57
] have suggested that white matter atrophy occurs in later stages of frontotemporal dementia and reflects severe gray matter and neuronal cell loss, but other reports have demonstrated that white matter atrophy occurs during earlier stages of the disease [19
]. Furthermore, recent studies have reported change in white matter integrity of patients with frontotemporal dementia using diffusion tensor imaging markers [20
]. Our findings support and extend existing literature suggesting that white matter degeneration may be a direct consequence and a major pathological feature of bvFTD, consistent with histopathological evidence of tau deposition in white matter of bvFTD patients [58
Myelination and the maintenance and repair of these highly vulnerable tissues is responsible for the exquisitely synchronized timing of action potentials that underlie neuronal network oscillations necessary for optimal human cognitive and behavioral functioning [59
]. Regions such as the frontal and temporal lobes that myelinate later in the developmental process have fewer oligodendrocytes supporting greater numbers of axons compared to earlier-myelinated regions (e.g., primary sensorimotor regions), and the axons are smaller with fewer myelin lamellae [24
]. Oligodendrocytes in these cortical association areas have particularly high metabolic demands in order to maintain the widely distributed network of axons, which makes these axons more susceptible to breakdown from pathological processes [59
]. The symptoms most often manifested by patients with FTLD (behavioral inhibition, language disturbance) may result from disruption of the circuitry containing late-myelinating, high-workload fibers and eventual white matter degradation in the cortical association regions [60
Another novel finding in our study is that the pattern of atrophy exhibited by our NF-PPA group was more global than expected, involving frontal, temporal, and parietal regions, and they did not demonstrate the characteristic left-dominant pattern associated with this population [61
]. Even though the bvFTD group exhibited greater atrophy in the VMPFC than NF-PPA participants, the NF-PPA group displayed a significantly greater rate of atrophy than either bvFTD or SV-PPA groups in the posterior frontal and parietal lobes, particularly in the white matter, further highlighting the involvement of white matter in the pathology of FTLD. As the disease progresses, atrophy in NF-PPA can extend posteriorly along the Sylvian fissure into the parietal lobe [62
]. However, we also acknowledge the possibility that the observed increase in parietal atrophy may indicate underlying Alzheimer's pathology in some of the NF-PPA subjects. Several reports have documented cases of autopsy-confirmed Alzheimer's pathology in patients who presented with clinical symptoms of NF-PPA [63
]. Alternatively, a subset of patients with nonfluent aphasia has been documented to progress into cortico-basal degeneration, which share a common pathology of tau inclusion and exhibit frontoparietal lobe atrophy on neuroimaging [70
]. Therefore, our NF-PPA group may represent an admixture of heterogeneous etiology, which can only be verified through pathological studies.
Our results revealed that specific behavioral symptoms at baseline predicted longitudinal changes in corresponding brain regions. The presence of frontal lobe-mediated behavioral manifestations, such as euphoria, apathy, and disinhibition, was associated with greater longitudinal volume loss of the VMPFC. Measures that assess a broader spectrum of both ‘positive’ and ‘negative’ frontal-executive symptoms were sensitive to rates of atrophy in frontal lobes and white matter volume. Similar structural-function relationships were observed within SV-PPA and NF-PPA subgroups when examined separately (data not shown). These findings underscore the specificity in the relationship between focal anatomical abnormalities and behavioral expression in this population, which traverses the boundaries of putative clinical definitions. Examination of the relationship between prospective changes in brain volume with deterioration in clinical and cognitive symptoms yielded only a marginally significant correlation between change in EXEC score and total white matter volume but not with more specific regions. The absence of more striking associations may be partially attributed to floor effects as performance on select measures may already be severely impaired at baseline for specific syndromic groups (i.e., BNT for SV-PPA), which then obscures the magnitude of the change scores.
The strengths of the present study include the longitudinal design in which each subject acts as his/her own control; measurement of intraindividual rates of change yields greater sensitivity for detection of subtle brain changes over time and overcomes the problem of interindividual variation in brain structure. Several important study limitations should also be acknowledged. Combining scans from multiple centers increases the power of the study but introduces the possibility of intercenter variability in diagnosis, MR instrument, and procedures. However, the participating centers are known for their expertise in FTLD, and the study design is balanced such that each site contributed participants from each clinical group, thus scanner differences affect all subjects and is not specific to one clinical subgroup. Another limitation is the uncertainty of the definitive diagnosis or the underlying histopathology of the patients studied. Careful follow-up of patients without a change in their clinical diagnoses may improve the gold standard in clinical studies. The SV-PPA group was significantly older than the bvFTD group, but adjusting for age did not meaningfully alter the results. Furthermore, one would expect increased age to be associated with greater rates of atrophy, but the younger bvFTD group actually exhibited more atrophy than SV-PPA.
The ability to track regional atrophy over time can help refine differential diagnosis and monitor disease progression in FTLD and identify patterns of neurodegeneration specific to each clinical syndrome. Even though the subjects were scanned only 1 year apart, our TBM data were able to show distinct regional patterns of atrophy, suggesting that this method can be a sensitive approach for detecting syndrome-specific longitudinal brain changes. Additionally, significantly greater progression of white matter loss was observed in the bvFTD group compared to the semantic-variant syndromes of FTLD, which was also related to frontal lobe behavioral symptoms. Even though there are limitations to our study as the findings in the NF-PPA group were less specific, further investigation of early changes in white matter would be merited given that it may provide potentially sensitive indicators for detecting syndrome-specific longitudinal brain changes.