Tau imaging detects distinctive distribution of tau pathology in ALS/PDC on the Kii Peninsula.

OBJECTIVE: To characterize the distribution of tau pathology in patients with amyotrophic lateral sclerosis/parkinsonism dementia complex on the Kii Peninsula (Kii ALS/PDC) by tau PET using [11C]PBB3 as ligand.
METHODS: This is a cross-sectional study of 5 patients with ALS/PDC and one asymptomatic participant with a dense family history of ALS/PDC from the Kii Peninsula who took part in this study. All were men, and their age was 76 ± 8 (mean ± SD) years. Thirteen healthy men (69 ± 6 years) participated as healthy controls (HCs). Dynamic PET scans were performed following injection of [11C]PBB3, and parametric PET images were generated by voxel-by-voxel calculation of binding potential (BP* ND) using a multilinear reference tissue model. [11C] Pittsburgh compound B (PiB) PET, MRI, and cognitive tests were also performed.
RESULTS: A voxel-based comparison of [11C]PBB3 BP* ND illustrated PET-detectable tau deposition in the cerebral cortex and white matter, and pontine basis including the corticospinal tract in Kii ALS/PDC patients compared with HCs (uncorrected p < 0.05). Group-wise volume of interest analysis of [11C]PBB3 BP* ND images showed increased BP* ND in the hippocampus and in frontal and parietal white matters of Kii ALS/PDC patients relative to HCs (p < 0.05, Holm-Sidak multiple comparisons test). BP* ND in frontal, temporal, and parietal gray matters correlated with Mini-Mental State Examination scores in Kii ALS/PDC patients (p < 0.05). All Kii ALS/PDC patients were negative for [11C]PiB (β-amyloid) except one with marginal positivity.
CONCLUSION: [11C]PBB3 PET visualized the characteristic topography of tau pathology in Kii ALS/PDC, corresponding to clinical phenotypes of this disease.

The Kii Peninsula of Japan is a high-incidence foci of amyotrophic lateral sclerosis/parkinsonism dementia complex (ALS/PDC), similar to the island of Guam.[1] ALS and PDC clinically occur separately or in combination, and most investigators consider ALS and PDC to be different manifestations of a single disease entity.[1-3] The neuropathologic hallmarks of ALS/PDC of Kii are widespread neurofibrillary tangles (NFTs) and neuropil threads, most predominantly in medial temporal and frontal cortices, and less in other cortices, subcortical nuclei, brainstem, and spinal cord.[1-3] Tau-positive astrocytes were also focally present in white matter.[3] NFTs in Kii ALS/PDC are ultrastructurally characterized as helical filaments composed of all 6 tau isoforms, similar to those in Alzheimer disease (AD) and Guamanian ALS/PDC.[1,4,5] Kii ALS/PDC differs from AD because of different NFT distribution and lack of abundant senile plaque in brain.[1-3]

Genetic and environmental factors are implicated in Kii ALS/PDC pathogenesis.[6] The high-incidence foci of ALS/PDC on the Kii Peninsula may provide a clue to the pathogenesis of this and other related neurodegenerative disorders. It would be particularly important to clarify the amount and spatial extent of tau pathology among living Kii patients with ALS/PDC to identify modifiers of tau fibril formation and dissemination.

We developed 2-([1E,3E]-4-[6-([11C]methylamino)pyridin-3-yl]buta-1,3-dienyl)benzo[d]thiazol-6-ol ([11C]PBB3) as a PET imaging agent for pathologic tau lesions, and have successfully imaged tau aggregates in patients with AD and a patient with corticobasal syndrome by [11C]PBB3 PET.[7,8] Based on this in vivo imaging technology, the present study aims to characterize the distribution of tau aggregates in relation to clinical phenotypes of Kii ALS/PDC.

Methods

Five patients with ALS/PDC[9] and one asymptomatic participant with a dense family history of ALS/PDC from the Kii Peninsula took part in this study. Thirteen healthy men (69 ± 6 years old) participated as healthy controls (old HCs). All participants underwent neuropsychological assessments, MRI, and PET scans with [11C]PBB3 for tau imaging, and with [11C] Pittsburgh compound B (PiB) for β-amyloid (Aβ) imaging. Details of the methods are presented in appendix 1.

Results

Demographics

Demographics of Kii ALS/PDC patients and old HCs are summarized in the table. Six Kii ALS/PD patients consisted of 1 asymptomatic participant, 4 ALS/PDC patients, and 1 PDC patient. There was no pure ALS patient. There was no notable age difference between the 6 Kii ALS/PD patients and the 13 old HCs. Years of education of Kii ALS/PDC patients were less than those of old HCs (p < 0.01). Kii ALS/PDC patients presented lower Mini-Mental State Examination (MMSE)[10] and Frontal Assessment Battery (FAB) scores,[11] and higher Clinical Dementia Rating (CDR) sum of boxes (SOB),[12] Neuropsychological Inventory (NPI),[13] and Unified Parkinson's Disease Rating Scale (UPDRS) motor scores,[14] than old HCs (p < 0.01).

Table

Demographics of Kii ALS/PDC patients and old HCs

PET imaging

[11C]PBB3 binding was increased in characteristic brain regions of Kii ALS/PDC patients compared with old HCs (figures 1–4). SPM12 (Wellcome Department of Cognitive Neurology, London, UK) analysis of the asymptomatic participant (patient 1) vs the 13 old HCs showed regions with increased [11C]PBB3 binding potential (BP*ND) in cerebral gray and subcortical white matter and the pontine base of this individual (figure 2). In patient 2, [11C]PBB3 BP*ND was increased in the deep cerebral white matter and brainstem, including the corticospinal tract,[15] rather than neocortical gray matter areas. Noticeable increases of [11C]PBB3 BP*ND were more extensively observed in cerebral gray and white matter regions, brainstem, and cerebellum of patients 3, 5, and 6. There was a trend of increased BP*ND in the brain associated with higher CDR SOB, but patient 4 was an outlier with only modest increase of [11C]PBB3 BP*ND in the brain compared with patients 3, 5, and 6. High BP*ND was additionally seen in the striatum of patients 3 and 6 (figure 2). Group-wise 2-sample t test by SPM demonstrated increased [11C]PBB3 BP*ND in cerebral gray and subcortical white matter, deep white matter including the corticospinal tract, pontine basis, and cerebellum of Kii ALS/PDC patients compared with old HCs (figure 3).

Figure 2

Increase of [11C]PBB3 binding potential (BP*ND) in each of the Kii amyotrophic lateral sclerosis/parkinsonism dementia complex patients compared with old healthy controls

Two axial images (z = 76, 108), a coronal image (y = 120), and a sagittal image (x = 91) were displayed on an MRI template (ch2bet.nii.gz in MRIcron) in each patient. Yellow indicates voxels at p < 0.05 (uncorrected, >50 voxels) and red indicates voxels at p < 0.01 (uncorrected, >50 voxels).[11C]PBB3 = 2-([1E,3E]-4-[6-([11C]methylamino)pyridin-3-yl]buta-1,3-dienyl)benzo[d]thiazol-6-ol; CDR = Clinical Dementia Rating; SOB = sum of boxes.

Figure 3

Increase of [11C]PBB3 binding potential (BP*ND) in 6 Kii amyotrophic lateral sclerosis (ALS)/parkinsonism dementia complex (PDC) patients compared with 13 old healthy controls (HCs)

Seven trans-axial slices (z = 43, 55, 67, 79, 91, 103, 115 mm) are displayed. Yellow indicates voxels at p < 0.05 (uncorrected, >50 voxels) and red indicates voxels at p < 0.01 (uncorrected, >50 voxels).[11C]PBB3 = 2-([1E,3E]-4-[6-([11C]methylamino)pyridin-3-yl]buta-1,3-dienyl)benzo[d]thiazol-6-ol.

Representative [11C]PBB3 binding potential (BP*ND) PET images

PET images in the upper row are of a 78-year-old Kii amyotrophic lateral sclerosis (ALS)/parkinsonism dementia complex (PDC) patient (case 4) and those in the lower row are of a 76-year-old healthy control (HC). Two trans-axial and one coronal PET image are displayed in the upper and lower rows. Extensive regions with high BP*ND are seen in the cerebral cortex in the Kii ALS/PDC patient compared with HC. There are voxels with high BP*ND in the superior sagittal sinuses of the Kii ALS/PDC patient and the HC, which are thought to be off-target binding to venous sinuses.[7][11C]PBB3 = 2-([1E,3E]-4-[6-([11C]methylamino)pyridin-3-yl]buta-1,3-dienyl)benzo[d]thiazol-6-ol.

In a group-wise volume of interest (VOI) analysis of [11C]PBB3 PET data, BP*ND for this radioligand was higher in the hippocampus and in frontal and parietal white matter of Kii ALS/PDC patients than in old HCs (figure 4). In Kii ALS/PDC patients, MMSE scores were positively correlated with BP*ND in frontal (rs = −0.886, p < 0.05), temporal (rs = −0.943, p < 0.05), and parietal gray matter (rs = 0.943, p < 0.05) by Spearman correlational analysis. NPI scores were also correlated with BP*ND in frontal gray matter (rs = 0.886, p < 0.05). There was no correlation between CDR-SOB, FAB scores, UPDRS motor scores, and BP*ND in any regions of these patients.

Figure 4

Regional binding potential (BP*ND) of [11C]PBB3 in 13 old healthy controls (HCs) and 6 Kii amyotrophic lateral sclerosis (ALS)/parkinsonism dementia complex (PDC) patients

*p < 0.05, **p < 0.005 by Holm-Sidak multiple comparisons test. Regions highlighted in bold type indicate those with differences in BP*ND between old HCs and Kii ALS/PDC patients.[11C]PBB3 = 2-([1E,3E]-4-[6-([11C]methylamino)pyridin-3-yl]buta-1,3-dienyl)benzo[d]thiazol-6-ol; GM = gray matter; WM = white matter.

[11C]PiB PET images were negative in the 6 Kii ALS/PDC patients except for patient 4, who showed marginal positivity. All HCs were PiB-negative, and there was no difference in [11C]PiB between Kii ALS/PDC patients and old HCs (table).

T1-weighted MRI of patient 1 was normal. T1-weighted MRI of patients 2–6 showed mild to moderate frontal, temporal, and parietal atrophy (figure 5).

Figure 5

Brain T1-weighted MRI of Kii amyotrophic lateral sclerosis/parkinsonism dementia complex patients

Kii ALS/PDC brain tissue staining

Numerous NFTs and neuropil threads in the CA1 sector of the hippocampus were found to be positive for PBB3 and Gallyas-Braak silver staining in a histochemical analysis of the first series brain sections derived from 3 Kii ALS/PDC patients (CA1 of 65M is shown in figure 6A). However, the majority of NFTs in this area were negative for AT8, and accordingly were conceived to be extracellular ghost tangles.[16,17] Moreover, tau aggregates in neuronal somas and putative axons were observed in the motor cortex as inclusions triply labeled with PBB3, AT8, and Gallyas-Braak silver staining (pathologies in the motor cortex of 63F shown in figure 6B). Axonal threads were triply stained in white matter adjacent to the cortex of 2 patients (77M and 63F; figure 6, C and D). Immunoreactive tau aggregates in neuronal somas and neurites were also noted in the striatum (staining in the posterior dorsal putamen of 63F is shown in figure 6E), but most of these tau aggregates was only modestly labeled with PBB3 and Gallyas-Braak silver staining, suggesting a relatively low packing density of these fibrillary assemblies.[17]

Figure 6

PBB3 fluorescent, AT8, and Gallyas-Braak (GB) staining in postmortem brains with Kii amyotrophic lateral sclerosis (ALS)/parkinsonism dementia complex (PDC) in the first (A–E) and second (F–H) series

In the first series samples, numerous neurofibrillary tangles (NFTs) are positive for PBB3 and GB staining in the CA1 sector of the hippocampus, although the majority of NFTs in this area were negative for AT8 (A), in agreement with histopathologic features of extracellular ghost tangles. Tau aggregates in neuronal somas and putative axons were observed in the motor cortex as inclusions labeled with PBB3, AT8, and GB staining (B). Axonal threads in white matter adjacent to the cortex of 2 patients are labeled with PBB3, AT8, and GB staining (C, D). In the posterior dorsal putamen, there are tau aggregates in neuronal somas and neurites clearly labeled with AT8 but only partly labeled with PBB3 and GB staining (E), indicating the presence of tau deposits with relatively lower packing densities similar to pretangles. In the second series samples, astrocytic plaque-like inclusions are labeled with PBB3, AT8, and GB staining in the premotor cortex (F). Axonal threads are labeled with PBB3, AT8, and GB staining in white matter adjacent to the premotor cortex (G). Tau aggregates resembling oligodendrocytic coiled bodies beside axonal threads are labeled with PBB3, AT8, and GB staining in the posterior limb of the internal capsule (white matter) (H). Red arrowheads indicate tau aggregates. [11C]PBB3 = 2-([1E,3E]-4-[6-([11C]methylamino)pyridin-3-yl]buta-1,3-dienyl)benzo[d]thiazol-6-ol.

In the second series of specimens, PBB3-, AT8-, and Gallyas-Braak-positive tau aggregates were found as astrocytic plaque-like inclusions in gray matter (pathologies in the premotor cortex of 71F is shown in figure 6F) and axonal threads in white matter (pathologies in subcortical white matter of 70F is shown in figure 6G). Triple labeling of tau pathologies was also detected as lesions resembling oligodendrocytic coiled bodies beside axonal threads in white matter (staining in the posterior limb of the internal capsule of 71F is shown in figure 6H). These findings suggest that increased radioactivity retention in [11C]PBB3 PET is attributable to radioligand binding to 4-repeat tau deposits in gray and white matter of a small subset of Kii ALS/PDC patients.

Discussion

[11C]PBB3 PET demonstrated increased radioligand BP*ND, which is indicative of tau accumulation, in the brain of all 6 Kii ALS/PDC patients compared with 13 old HCs. Even in the asymptomatic patient (patient 1), [11C]PBB3 binding was enhanced in gray matter of the cerebral cortex including the medial temporal lobe, subcortical white matter, and pontine base. Interestingly, a Kii ALS/PDC patient (patient 2), who exhibited prominent upper motor neuron signs, had increased tau accumulation in deep white matter and pontine base containing the corticospinal tract. There were more widespread regions with increased tau accumulation involving cortical and subcortical gray matter, white matter, and brainstem in Kii ALS/PDC patients with dementia (patients 3, 5, and 6). These findings suggest intimate associations between the topology of tau depositions and impairments of regional functions in the Kii ALS/PDC brains, implying that tau-induced neurotoxicity leads to deterioration of local neurons. This notion is supported by the present neuropathologic examination demonstrating abundant ghost tangles in the hippocampus of the patients.

There was a trend of increased tau accumulation in the brains of Kii ALS/PDC patients in association with CDR-SOB. However, patient 4 was an outlier with only modest increase of tau accumulation in the brain, and was also marginally positive for Aβ. The patient was born in the south part of the Kii peninsula, moved elsewhere at age 3, and developed symptoms 73 years later.[18] Relatively low tau accumulation in the brain of this patient despite CDR-SOB 3 could be attributed to the shortest stay in the endemic area, and the longest incubation to develop the disease, and functional deficits in the brain of this patient may also result from accumulation of other pathogenetic proteins including Aβ.

The distribution of tau accumulation was rather scattered and variable among Kii ALS/PDC patients. A previous neuropathologic study showed that NFTs tended to appear discontinuously and nonsystematically in Kii ALS/PDC.[3] This was in contrast to the neuropathologic features of AD, in which NFTs are confined to gray matter, and spread systemically according to Braak histopathologic stages.[7,8,19] VOI analysis of [11C]PBB3 PET showed that tau aggregates were preferentially accumulated in the hippocampus and in frontal and parietal white matter in the 6 Kii ALS/PDC patients compared with the 13 old HCs, although voxel-wise SPM analysis did not reveal tau accumulation in the hippocampus of these patients. This discrepancy between VOI and voxel-wise analyses might be explained by the diversity of the subregional distributions of tau aggregates in the hippocampus among the 6 patients.

Previous pathologic studies established that NFTs and neuropil threads are the main pathology of Kii ALS/PDC.[1-3] In addition, tau-positive astrocytes are also found predominantly in subpial and perivascular areas and are focally seen in the white matter of all patients with Kii ALS/PDC.[3] Our brain tissue staining study demonstrated that tau-positive threads in white matter are stained with PBB3 in all examined patients, although tau-positive astrocytes are stained with PBB3 only in a subpopulation of Kii ALS/PDC patients with corticobasal degeneration–like findings such as depositions resembling astrocytic plaques. The present PET study showed high [11C]PBB3 binding in white matter of the Kii ALS/PDC patients by VOI analysis. Voxel-based analysis of [11C]PBB3 parametric images showed increased [11C]PBB3 binding in not only subcortical white matter but also in deep white matter. Taken together, high [11C]PBB3 binding in white matter should at least partly reflect tau pathologies in the white matter of Kii ALS/PDC patients, although it could also be attributed to the tau pathologies in cortical gray matter around the depth of a sulcus and gray–white boundary in the present image analysis, due to relatively low spatial resolution of PET data. Our preliminary histochemical assessments of brain sections indicated that the density of tau aggregates in gray matter in the vicinity of white matter is more than 10-fold higher than the density of tau-positive threads in white matter. Hence, [11C]PBB3-PET signals in superficial white matter might be largely affected by tau deposits in deep gray matter. Another potential technical issue may be the possible binding of the radioligand to off-target components in white matter, as exemplified by cross-reactivity of THK-5,351 with monoamine oxidase B (MAO-B) expressed in activated astrocytes.[20] However, our recent data have proven that PBB3 does not bind to MAO-B.[21] Hence, PET imaging with [11C]PBB3 could offer an in vivo technology to capture AD-like and CBD-like heterogeneous tau pathologies in Kii ALS/PDC.

The present results suggest that increased tau accumulation in gray matter of the frontal, temporal, and parietal lobes leads to cognitive deficits in Kii ALS/PDC, as assessed by MMSE. There was a correlation between tau accumulation in frontal gray matter and NPI scores, suggesting that tau accumulation also may contribute to neuropsychiatric symptoms such as apathy and irritability in Kii ALS/PDC patients.

Neuropathologic studies of Kii ALS/PDC patients showed that there were various types of phosphorylated α-synuclein-positive lesions including neuronal cytoplasmic inclusions, dystrophic neurites, and glial cytoplasmic inclusions.[3,22] Phosphorylated α-synuclein was distributed mainly in the limbic system and brainstem of Kii ALS/PDC brains,[3,22] while tau pathology is more prevalent than α-synuclein pathology in most affected areas.[22] Furthermore, our previous study demonstrated that PBB3 binding to α-synuclein deposits is 10–50 times less than to tau deposits.[23] Therefore, the increase of [11C]PBB3 binding in the present study was mostly attributable to tau pathology in Kii ALS/PDC patients.

Only 2 (patients 3 and 6) of 5 Kii ALS/PDC patients with parkinsonism showed high radioligand binding in the striatum and midbrain including the substantia nigra in comparison with old HCs. Parkinsonism could result from neuronal loss in the substantia nigra induced by tau aggregates with or without α-synuclein deposits in these Kii ALS/PDC patients, as Lewy body pathology is often found in the substantia nigra of Kii ALS/PDC brains.[3] However, it should be noted that Lewy bodies, and Lewy neurites in dementia with Lewy bodies, are thought to be undetectable in vivo by [11C]PBB3 PET according to our previous data.[23]

In addition to tau and α-synuclein depositions, TDP-43 pathology is noted in the brain and spinal cord of patients with Kii ALS/PDC, similar to sporadic ALS.[3] It has been reported that a patient with progressive supranuclear palsy phenotype associated with DCTN1 mutation showed increased [11C]PBB3 binding in basal ganglia and parietal lobe.[24] Mutations of the DCTN1 gene have been mostly associated with ALS and Perry syndrome, and one of the neuropathologic features of Perry syndrome is the accumulation of ubiquitinated TDP-43-positive neuronal inclusions, dystrophic neurites, glial cytoplasmic inclusions, and axonal spheroids. However, there have been no reports documenting the histopathology of progressive supranuclear palsy phenotype associated with DCTN1 mutation, and therefore the cross-reactivity of [11C]PBB3 with TDP-43 lesions remains to be determined.

There are a few technical considerations related to the quantification of radioligand binding in this study. SUVR, which is a target to reference the tissue radioactivity ratio at a near-equilibrium state of tracer kinetics, is usually used as an in vivo index of amyloid and tau accumulations. We calculated BP*ND of [11C]PBB3 in the kinetic model instead of with SUVR,[25] although this requires dynamic PET scans for several 10-minute periods. The advantage of BP*ND over SUVR as an index of radioligand binding is that BP*ND is minimally affected by changes in cerebral blood flow. Moreover, the cerebellum is usually used as reference tissue, because this area lacks amyloid and tau pathologies in AD. In contrast, a recent neuropathologic study showed that phosphorylated tau pathology of various types is present in the dentate nucleus and Purkinje and glial cells in the cerebellum of patients with Kii ALS/PDC.[26] Therefore, we used gray matter voxels, which have a low likelihood of bearing tau pathology, as a reference instead of the cerebellum.[27] BP*ND measures in white matter might be biased due to a possible difference in nondisplaceable binding of PBB3 between gray matter and white matter. However, the current reference tissue determined in gray matter does not provoke a pronounced bias affecting interpretation of the results when comparing the patient and control groups, since the bias at issue should be constant among Kii ALS/PDC patients and old HCs and should not cause larger differences in white matter regions relative to gray matter regions. An alternative option to quantify specific binding of [11C]PBB3 in white matter would be to generate reference tissue data by extracting voxels from white matter only on the basis of a white matter frequency histogram. However, the signal-to-noise ratio in white matter reference tissue may not be as high as the ratio in gray matter reference tissue, due to low nondisplaceable radioligand binding in white matter relative to gray matter. Accordingly, this would lead to decreased accuracy in the quantification of tau pathologies in white matter.

The limitation of this study is that only a small number of male Kii ALS/PDC patients were included in the PET imaging assays. Since the occurrence of Kii ALS/PDC is not sex-specific, PET studies with a large number of Kii ALS/PDC patients including female patients will be required to further validate the current findings. In addition, a pursuit of the relationships between neuroimaging and neuropathologic data in the same patients with Kii ALS/PDC will ascertain the utility of [11C]PBB3 PET for assessing tau lesions in living patients.

[11C]PBB3 PET visualized the characteristic topography of tau pathology in Kii ALS/PDC, corresponding to clinical phenotypes of this disease.