In this study we compared clinical features, fibrillar Aβ deposition, and regional glucose metabolism in matched patients presenting clinically with PCA and AD. As expected, patients with PCA showed greater impairment in visuospatial tasks and patients with AD showed relative deficits in episodic memory. All patients with PCA showed high PiB uptake, consistent with previous reports that the clinical syndrome of PCA is very often associated with underlying AD pathology.1–3
The pattern of cortical PiB binding in PCA was diffuse, affecting anterior and posterior cortical regions, visual and nonvisual areas, and on direct statistical comparison was indistinguishable from the pattern seen in AD. In contrast, patients with PCA showed a more posterior pattern of glucose hypometabolism with greater occipital involvement than was seen in patients with clinically “typical” AD. These findings suggest that distinct clinical features and regional hypometabolism in PCA and AD are not related to fibrillar amyloid distribution.
Previous autopsy studies have yielded conflicting results regarding the distribution of amyloid pathology in PCA, with some studies reporting 3–5 times more plaques in visual areas in patients with PCA compared to patients with typical AD,5
and others finding similar plaque counts in these regions in PCA and AD.2,3
The reasons for these discrepant findings are unclear, but may include subject factors (e.g., age, disease severity, failure to control for copathology) as well as the methods used to quantify plaques (e.g., lesion counts vs more rigorous quantification, staining techniques, discrimination between diffuse vs neuritic plaques). Our data suggest that, in the mild-to-moderate disease stage, there is no difference in the distribution of fibrillar Aβ pathology between PCA and AD. These findings are congruent with observations from patients with underlying AD who present with another focal cortical syndrome, primary progressive aphasia (PPA). Though neurodegeneration in PPA is highly asymmetric and preferentially involves the language network,30,31
the distribution of amyloid pathology (as measured by PiB or at autopsy) in AD-associated cases is symmetric and indistinguishable from that found in typical AD.14,32
Previously, our group used voxel-based morphometry (VBM) to compare atrophy patterns in patients with early age-at-onset AD, PCA, and logopenic aphasia (LPA), a PPA variant associated with underlying AD.31
We found that patients with all 3 syndromes showed overlapping atrophy in precuneus/posterior cingulate and lateral temporoparietal cortex. Additional right ventral-occipital/parietal atrophy was found in PCA and left middle/superior temporal gyrus atrophy in LPA. Similar results were reported in another study contrasting atrophy patterns in AD and PCA.31
Likewise, in this study we found comparable glucose hypometabolism in AD and PCA in temporoparietal cortex and precuneus, with extension of hypometabolism into occipitotemporal cortex in PCA. These observations suggest that precuneus/posterior cingulate and lateral temporoparietal involvement is a common feature of AD-associated syndromes, with extension of neurodegeneration into other, distinct regions in PCA and LPA. It is intriguing to speculate whether common involvement of these regions in AD is related to their proposed function as highly interconnected “cortical hubs,” which may render them susceptible to both early Aβ aggregation (due to high synaptic activity with resultant Aβ release) and Aβ-mediated neurodegeneration (due to high metabolic demand).33
The high level of connectivity may also enable the spread of disease from these regions into several cortical networks,34
including those underlying memory, language, and visual function. Our work in early-onset AD,15
and now PCA suggests that the respective involvement of these networks is not explained by the distribution of amyloid. Rather, it may be the relative vulnerability of networks in an individual that determines the neurodegenerative pattern and clinical phenotype of AD. In most patients the posterior “default mode network” may be most vulnerable to Aβ-mediated neurodegeneration, perhaps because of high metabolic demand,35
leading to a “typical” amnestic AD phenotype. In individuals with PPA, however, the language network may be particularly vulnerable, as suggested by the high rate of developmental language disorders in individuals who develop PPA later in life.36
To our knowledge, such premorbid risk factors have not been systematically studied in PCA, where visual networks are disproportionately affected. Future studies that estimate premorbid “reserve” in specific cognitive domains, investigate the integrity and function of the associated neural networks, and relate these findings to clinical and degenerative phenotype may help elucidate the mechanisms of selective network degeneration in focal variants of AD.
Another possible explanation for the dissociation between the patterns of PiB and FDG binding in this study is that the fibrillar Aβ deposits imaged by PiB may not be the critical pathology driving neurodegeneration. Soluble Aβ oligomers are considered to be the most neurotoxic Aβ species,37,38
yet are not bound by PiB. It is possible the patients with PCA have high concentrations of Aβ oligomers in visual areas, though it is not clear why this would not be reflected by higher concentrations of fibrillar Aβ that is thought to be in equilibrium with the soluble compartment. Finally, the degenerative pattern is likely more closely related to the distribution of neurofibrillary pathology. Higher neurofibrillary tangle (NFT) counts in visual regions in PCA compared to AD is a consistent finding across pathology studies,2,3,5
and asymmetric NFT have also been reported in PPA.32
It is unknown, however, whether in AD NFT form independently or secondary to Aβ-driven processes.
Our results differ from case reports of single subjects with PCA with posterior-predominant PiB binding10
or disproportionate occipital PiB compared to AD.9
While most subjects with PCA in our cohort showed diffuse PiB binding, individual subjects with posterior-predominant binding patterns could be identified, though conversely frontal-predominant binding was seen in one subject, and similar variability in binding patterns was also seen in the “typical” AD group. Following atrophy correction, there was a trend toward higher mean PiB values for PCA compared to AD in the cuneus and lingual gyrus (p
= 0.09), leaving open the possibility that subtle increases in occipital amyloid are found in PCA and that our study was underpowered to detect them, though it is doubtful that they are of the 3- to 5-fold higher magnitude reported in some pathology studies.5
Conversely, this result may be due to overinflation of PET counts by the atrophy correction procedure, which remains controversial for PiB data. Importantly, regions that showed marked hypometabolism in PCA compared to AD showed essentially identical PiB uptake in the 2 groups even after atrophy correction (e.g., lateral occipitotemporal and inferior occipital cortex), supporting the assertion that PiB and FDG patterns are dissociated in the 2 syndromes.
Our study has limitations. The sample sizes were relatively small, potentially limiting our power to detect differences between AD and PCA, though the size of our cohorts is comparable to those found in previous studies of PCA, and had sufficient power to detect differences in FDG binding and marginal differences in clinical performance. Although initial studies suggest that in vivo PiB binding correlates highly with postmortem Aβ measures,11,8
the limitations of this technique have not yet been fully identified. It is possible that patients with PCA in our study had a secondary pathology in addition to AD, though this is less likely given the exclusion of subjects with clinical features suggestive of dementia with Lewy bodies, corticobasal degeneration, or prion disease. Since we matched the AD and PCA groups for age, our AD control group is largely composed of early age-at-onset patients (mean age at onset 58.8 years). This may have minimized both clinical and anatomic differences between AD and PCA, since patients with early-onset AD have greater visuospatial impairment and posterior cortical atrophy and relatively preserved episodic memory and medial temporal lobes compared to late-onset patients.39,40