Primary Age-Related Tauopathy (PART): Addressing the Spectrum of Neuronal Tauopathic Changes in the Aging Brain

Abstract

Purpose of Review

Primary age-related tauopathy (PART) was recently proposed as a pathologic diagnosis for brains that harbor neurofibrillary tangles (Braak stage ≤ 4) with little, if any, amyloid burden. We sought to review the clinicopathologic findings related to PART.

Recent Findings

Most adult human brains show at least focal tauopathic changes, and the majority of individuals with PART do not progress to dementia. Older age and cognitive impairment correlate with increased Braak stage, and multivariate analyses suggest that the rate of cognitive decline is less than matched patients with Alzheimer disease (AD). It remains unclear whether PART is a distinct tauopathic entity separate from AD or rather represents an earlier histologic stage of AD.

Summary

Cognitive decline in PART is usually milder than AD and correlates with tauopathic burden. Biomarker and ligand-based radiologic studies will be important to define PART antemortem and prospectively follow its natural history.

Introduction

In 2008, den Dunnen and colleagues described the extraordinary neuropathologic findings of a 115-year-old woman who died of metastatic gastric adenocarcinoma [1]. Despite her advanced age, the brain showed no evidence of neurodegenerative disease nor cerebrovascular injury, but instead displayed neurofibrillary tangles (NFT) in the brainstem, amygdala, and transentorhinal and entorhinal cortices with minimal neocortical involvement. At nearly 114 years of age, her mini-mental state examination (MMSE) scored 26/30 and her remaining neuropsychiatric tests revealed performances that were superior to many 60- to 75-year-olds. The authors’ conclusions were that cognitive functions may extend beyond the usual lifespan and that neurodegenerative diseases of the brain are not inevitable.

This supercentenarian was exceptional, but while she was fortunate to evade neurodegeneration despite her longevity, the burden of dementia for aging societies is continuing to increase [2–4]. Examination of brains over the age of 75 years often reveals overlapping pathologic findings with that of neurodegenerative disorders, yet a subset of these individuals never manifest a neurological disorder [5–8]. Separating the usual histologic findings of healthy brain aging from that which are clinicopathologic can therefore be challenging, and this is exemplified in centenarian neuropathological studies that reveal a vast gamut of pathological hallmarks of various neurodegenerative diseases despite having apparently normal cognition [9–11].

One neuropathologic finding that intersects between healthy aging and dementia is the NFT. NFT are filamentous inclusions that are found within neurons and are comprised of aggregates of hyperphosphorylated tau (p-tau) [12, 13]. Comprehensive histopathologic studies of post-mortem brains reveal p-tau aggregates in the majority of adult brains, being almost always detected within the locus coeruleus [14, 15••, 16–19]. Furthermore, aggregation of p-tau is not a feature of a specific neurodegenerative disorder but can be found in ganglionic tumors, post-infectious states (e.g., subacute sclerosing panencephalitis (SSPE)), metabolic diseases such as Niemann–Pick C disease, and following trauma, as in chronic traumatic encephalopathy (CTE) [20–24]. Aggregates of p-tau are therefore common, increase with age, and, as an isolated biochemical finding, are somewhat non-specific to an etiology. Thus, accurate diagnosis and understanding of the disease process benefits from the clinical context as well as recognition of which cell types are affected by the tauopathy and the regional distribution of tauopathic burden within the brain.

For decades, it had been realized that there were many individuals with a clinical diagnosis of AD, yet at post-mortem examination, tauopathic changes were seen with minimal, if any, neuritic plaques [25–27]. In the absence of neuritic plaques, these brains could not be diagnosed as AD, and so to fill this diagnostic void, the terms “tangle-only dementia,” “neurofibrillary predominant dementia,” “senile dementia of the neurofibrillary tangle type,” and “senile dementia of the tangle type” were used [26–30]. Confusion then arose when it was realized that there were many aged brains that also harbored a similar p-tau aggregation pattern but without significant cognitive decline. This was further compounded when it became recognized how frequent intermediate Braak stage tauopathic changes were in the absence of plaques in autopsy series, a frequency that has been suggested to be between 2 and 10% of examined brains [31••]. In light of these challenges and to draw attention to this phenomenon, a group of experts in the field synthesized the literature and reached a consensus in formulating the concept of primary age-related tauopathy (PART) [31••]. This review explores the pathologic characteristics of PART and the recent developments of the clinical associations of this phenotype and discusses some of the controversies relating to the diagnostic entity.

Morphologic Features and Diagnostic Criteria

Currently, the gold standard method for the diagnosis of PART is post-mortem histological examination of the brain. Assignment of PART is independent of the antemortem cognitive status and consequently does not require knowledge of the presence or absence of dementia. Cases that were previously assigned as tangle-only dementia can therefore be designated as PART.

The macroscopic findings of PART are variable, but often no significant abnormalities are detectable in the brain for the individual’s age (Fig. 1a, b). In some instances, however, there may be visible mesial temporal lobe atrophy, and in fewer cases, there could be more widespread cerebral atrophy [28, 31••, 32, 33]. Microscopically, NFT are found in many of the same regions that are commonly afflicted in AD, but with minimal-to-absent involvement of the neocortex. Certain vulnerable subcortical nuclei are commonly involved with NFT, and attention is particularly drawn to the locus coeruleus, dorsal raphé nucleus, nucleus basalis of Meynert, and the amygdala [26]. Involvement of the cerebral cortex is concentrated within the medial temporal lobe where tangles are found in the parahippocampal gyrus, the ambient gyrus, and within the hippocampal sectors (Fig. 1d, i–k). The insular cortex is also affected. With this distribution in mind, the AD Braak staging system for NFT is used to document the extent of tauopathic changes in PART, and, hence, these brains range between Braak stages I–II (transentorhinal) and III–IV (limbic) [34]. Older individuals have higher Braak stages implying a stepwise progression of NFT in a similar fashion to AD [31••, 35••]. Within the hippocampal formation, both intracellular and extracellular (“ghost”) tangles may be found and, in severe cases, there may be spongiosis of the parahippocampal cortex [32]. However, the tauopathic burden in PART within the hippocampus appears to have a different pattern of distribution to that seen in classical AD. In contrast to AD, the CA2 subfield frequently harbors greater tauopathic burden than the adjacent CA1 sector in PART (Fig. 1e, f vs. m, n), which is a pattern that is also seen in CTE [26, 32, 36•, 37, 38]. Furthermore, there is a variable degree of tauopathic burden within the remaining hippocampal CA subfields, and the den-tate gyrus can be relatively spared [32].

Fig. 1.

Comparisons between definite PART of an 83-year-old woman (a–d, i–m) and AD neuropathologic changes (ADNC) [A3, B3, C3] of a 71-year-old man (e–h, n–r). a The brain was diffusely small with a whole fresh brain weight of 992.5 g. The left half brain shows no visible atrophy for age. b Likewise, a coronal section at the level of the lateral geniculate body shows no discernable cortical atrophy on cut section of the frontal, parietal, or temporal lobes. The subjacent white matter, thalamus, and hippocampal formation appears normal. c Luxol fast blue counterstained with hematoxylin and eosin (LH&E)–stained section shows no abnormalities in the hippocampal formation with a well-myelinated alveus and fimbria. d AT8 (p-tau) immunostained section of the hippocampal formation showing limited tauopathic burden within the parahippocampal gyrus (PHG) with less staining in the subiculum, CA2, and CA1 regions and the least labeling in the occipitotemporal gyrus (OTG). The boxes indicate the regions photographed in i–l. e In stark contrast to the brain with PART, this left half brain with AD shows severe frontotemporal atrophy with moderate parietal atrophy. f This frontotemporal atrophy is further shown in a coronal section of the left half brain at the level of the lateral geniculate body. There is white matter atrophy and widening of the temporal horn of the lateral ventricle that is associated with moderate hippocampal and mesial temporal atrophy. g An LH&E-stained section of the hippocampal formation shows atrophy and widening of the temporal horn of the lateral ventricle. h An AT8 immunostain shows widespread p-tau immunolabeling afflicting all hippocampal sectors and extending beyond the PHG to include the OTG. i, j More pyramidal neurons are tau-labeled in CA2 than in CA1 in this case of PART. k, l Many AT8-labeled neurons are seen in layer II of the parahippocampal gyrus (PHG) whereas few neurons (layer V shown) label in the OTG. m No plaques are seen in the supragranular layers of the OTG with a beta-amyloid immunostain in this case of PART. n, o Many tau-positive neurons and abundant neuropil threads are seen in CA2 (n) and CA1 (o) with more neurons labeling in CA1 than CA2 in this case of AD. p, q The tauopathic burden of AT8-labeled neurons and widespread neuropil threads are seen throughout the PHG (p) and the OTG (q). r Parenchymal amyloid plaques are highlighted by a beta-amyloid immunostain in layer II of the OTG of AD. Scale bar 50 μm

For a diagnosis of PART to be rendered, the amyloid burden should be low, if present at all, with a maximum Thal phase of 2. As would be expected with this prerequisite, cerebral amyloid angiopathy is infrequently seen [32]. A further distinction of PART is to discriminate brains with amyloid plaques from those without plaques by using the terms “possible PART” and “definite PART,” respectively (Table 1) [39]. In the original paper by Crary et al. and by others, it was recognized that as the Braak stage increased, the ratio of possible PART:definite PART increased and higher densities of amyloid plaques are seen with greater Braak stages [31••, 40••, 41]. For that reason, when pathologists are faced with the scenario of limbic Braak stages (III–IV) in the absence of amyloid plaques, one should consider FTLD-tau in their differential diagnosis, especially if the patient is in the younger age range [42].

Table 1.

Clinicopathologic designations of PART in relation to cognitive status and pathologic findings

Amyloid phase/clinical	No signs of cognitive decline	Cognitive decline/dementia
Thai 0	Asymptomatic, definite PART	Symptomatic, definite PART
Thai 1–2	Asymptomatic, possible PART	Symptomatic, possible PART

Other Associated Histologic Findings in PART

Since the histological findings of PART are strongly associated with aging, other age-related pathologies may coexist in these brains. Indeed, mixed brain pathologies are frequently seen in dementias [43–45]. Tau-immunoreactive astrocytes are sometimes found in PART and may comprise clusters of thorn-shaped astrocytes in gray/white matter (aging-related tau astrogliopathy (ARTAG)) or foci of tufted astrocytes/astrocytic plaques [32, 46]. Tau-labeled oligodendroglial inclusions (“coiled bodies”) have also been described [32]. When tufted astrocytes are widespread, the differential diagnosis should include progressive supranuclear palsy, or corticobasal degeneration when extensive astrocytic plaques are found.

Argyrophilic grains are also found with increasing frequency in the aging brain. One series found argyrophilic grain disease in 31% of brains with definite PART [35••, 47–49]. Small-to-moderate amounts of argyrophilic grains should not exclude the diagnosis of PART [31••].

Hippocampal sclerosis (HS) may be found in up to 10% of brains with PART, and this may be associated with phosphorylated TDP-43 pathology [32, 35••, 50]. As has been described in many neurodegenerative diseases and in aging, phosphorylated TDP-43 has also been reported in PART in approximately 30% of cases [35••, 42, 51–53]. This frequency is generally lower than that seen in AD (ranging from 36 to 74% of AD cases) [54–56]. The distribution of TDP-43-related changes mostly involves the limbic system, especially the amygdala, but may reach stage 3 of the TDP-43 staging system as proposed by Josephs et al. [35••, 57]. Another study reviewing TDP-43 abnormalities in PART described more extensive TDP-43 abnormalities with a few cases showing mild TDP-43 proteinopathy in the frontal and parietal neocortex [58].

While Lewy bodies (LB) are seen in PART, they are uncommon and seen at a frequency analogous to incidental LB of aging cohorts (6%) and at far lower occurrence that usually seen in AD (19–57%) [35••, 56, 59–61].

Finally, cerebrovascular disease is seen frequently in this aged cohort. Jellinger and Attems reported that over half of patients (54%) with neurofibrillary tangle predominant dementia harbored infarcts and were often striatal lacunes [32]. This frequency is comparable with that seen in other dementia cohorts [62–64].

Clinical Features

There are numerous clinical descriptions of the entity that is now referred to as PART, and a summary of the clinical findings from selected publications is provided in Table 2. Many of these accounts are largely consistent, whereas other reports, particularly of the more recent neuropsychiatric data, show conflicting data.

Table 2.

Summary of selected articles with clinical findings for individuals with a post-mortem diagnosis of PART

Article [reference]	Study design	Average age (years)	Sex (% women)	Summary of neuropsychiatric findings	Other
Reports prior to Crary et al. (2014) consensus paper
Ulrich, 1992 [27]	Retrospective clinicopathologic series of individuals with TOD (n = 10)	89.6	80	Five of the women were severely demented, two women suffered from a “psycho-organic syndrome,” one was disoriented and the two men experienced psychotic episodes at the end of life.	N/A
Bancher et al., 1994 [28]	Retrospective clinicopathologic study of brains with NTT but no amyloid plaque (n = 10)	89.6	80	Seven had a clinical diagnosis of probable AD; 3 with possible AD Two patients had paranoid symptoms/ depression Two had extrapyramidal signs
Ikeda et al., 1999 [30]	Retrospective clinicopathologic series of individuals with TOD (n = 4)	89	50	Gradual, progressive long-term memory loss, relative preservation of personality and cognitive function. Psychotic symptoms described in a subset of cases
Nelson et al., 2009 [65]	Retrospective clinicopathologic study (n = 26) with separate demographic data from the NACC registry (n = 219)	88.5 (n = 26)	46 (n = 26)	21 of the 26 individuals had normal cognition; 4 were diagnosed with AD and 1 with MCI Mean MMSE 26.5
Crary et al. (2014) consensus paper and reports since
Crary et al., 2014 [31••]	Review of the literature and consensus on PART	Mean range (Table 1 of paper) 77.0–92.0		Variable—some individuals may have amnestic symptoms
Josephs et al., 2017 [35••]	Retrospective clinicopathologic study of definite PART (n = 52); a subset with neuroimaging	88 at death (IQR 82–92)	56	Median MMSE 28/30 Relative sparing of episodic memory Poorer performances of executive functioning, cognitive speed, visuospatial abilities with increasing Braak stage	8% ApoE ε4 frequency Left anterior hippocampal atrophy
Jefterson-George et al., 2017 [66]	Retrospective clinicopathologic study of definite and possible PART from the NACC database (n = 226)	Median age ranged from 84 to 92	50	Poorer performance of semantic memory Poorer performance of cognitive speed and language ability with increasing Braak stage	N/A
Besser et al., 2017 [67]	Using the NACC registry to describe the clinicopathologic features of definite PART (n = 170) and cases with sparse amyloid (n = 207).	N/A	53	Definite PART participants were less frequently symptomatic than amyloid sparse individuals. Predictors of symptoms in the definite PART category included a history of depression, stroke and Braak stage.
Bell et al., 2018 [68]	Retrospective study comparing the clinicopathologic and genetic status of age-matched individuals with PART versus AD (n = 183)			Individuals with PART have a slower antemortem rate of decline in memory, language and visuospatial ability than patients with AD.	Frequency of ApoE ε4 is lower in PART (4.1%) than in AD (17.6%)
Besser et al., 2019 [41]	Retrospective comparison of neuropsychological test scores of individuals with PART (n = 426) and individuals with AD (n = 510) using the NACC registry	Majority were 80 and older	53	Definite PART had relative sparing of semantic memory and language compared with AD when matched by clinical dementia rating (CDR). When matched for CDR 0.5–1.0, definite PART had relative sparing of memory compared with AD. When matched for CDR 2.0–3.0, definite PART had relative sparing of attention. There were no significant differences in the cognitive domains between possible PART and AD	N/A
Teylan et al., 2019 [69]	Retrospective clinicopathologic study to assess clinical diagnosis of patients with cognitive impairment using NACC database of brains with PART (n = 161) and AD with moderate/ severe neuritic plaques (n = 1193).	83.1	42	Fewer patients with PART were diagnosed as AD in life compared with patients with AD. Only half of patients with dementia and PART were assigned a diagnosis of AD and were instead given a range of diagnoses including vascular brain injury (17.0%), primary progressive aphasia/bvFTD (12.5%)	Lower frequency of ApoE ε4 in PART than AD neuropathology
Teylan et al., 2020 [70•]	Retrospective clinicopathologic study of individuals with a baseline CDR 0.5 and comparing the rate of cognitive decline between PART (n = 126) and AD (n = 452) using NACC registry	86.1	48	Individuals with PART have slower cognitive decline than AD in all cognitive domains in a global composite score and with CDR-sum of boxes.	Frequency of ApoE ε4 is lower in PART (13.7%) than in AD (48.8%)

Before the 2014 consensus paper, the early descriptions of PART focused on patients with dementia, thus representing the more severe clinical aspect of the PART spectrum. In contrast to AD, these brains constituted a minority within dementia cohorts. The initial accounts portrayed a higher proportion of women being affected with an average age at death of 90 years old. The clinical phenotype of this dementia was mostly similar to AD, although psychotic episodes in a subset of individuals were also described [27, 30].

As it became steadily realized that these tauopathic changes were frequent in elderly brains without frank dementia, the PART consensus paper emerged to partly enable better recognition and consistent terminology of this histologic finding [31••, 65]. The result has been more clinicopathologic studies with larger sample sizes that incorporate data from neuropsychologic testing. Furthermore, several studies have utilized the clinical and autopsy data available in the National Alzheimer’s Coordinating Center’s (NACC) Database, information that has been curated from multiple Alzheimer Disease Research Centers across the USA [71].

In contrast to the older literature, recent studies of PART do not demonstrate an overwhelming gender predilection. As mentioned, older age does correlate with increasing Braak stage, and, to our knowledge, all of the PART studies have mean ages of death above 75 years. Subjective memory complaints (SMC) in this age group are common, and in a recent cohort of patients followed with SMC, the majority of patients with a post-mortem diagnosis of PART never progressed to dementia [72]. Furthermore, studies that have compared the rate of cognitive decline between PART and AD using multivariate analyses have consistently shown that the speed of deterioration is less than that of AD [68, 70•]. Two retrospective studies have demonstrated that patients with PART have poorer performance of semantic memory and that this deficit correlates with Braak stage [35••, 66]. The study by Josephs et al. reviewed brains with definite PART and argued that the deficit in semantic memory and relative sparing of episodic memory may be related to the anterior hippocampal atrophy that was seen in their cohort [35••]. In contrast, one other study that utilized the NACC database found that semantic memory was relatively preserved in definite PART when compared with AD and matched by clinical dementia rating (CDR) [41]. This discrepancy may be partly due to differences in study design. Josephs et al. was a study that utilized a standardized protocol and only included cases of definite PART, whereas Besser et al. utilized NACC data from different centers with differing protocols and assigned definite PART to cases without neuritic plaques. These cases may have inadvertently harbored some amyloid and therefore included possible PART cases. Interestingly, this latter study did not find significant cognitive differences between possible PART and AD suggesting clinical overlap between these two entities.

Patients with PART may have different clinical presentations from AD. Another study that utilized the NACC database found that just over half of post-mortem PART cases with dementia were clinically diagnosed as AD (51%) but at a much lower frequency than patients with AD who had dementia (85%). Instead, these patients with PART-dementia were issued other primary diagnoses such as vascular brain injury, behavioral variant of frontotemporal dementia (bvFTD), and primary progressive aphasia [69]. This diagnostic discrepancy may be a result of clinical differences between PART and AD [71].

Improved, reliable, biomarker studies that are less contingent on post-mortem neuropathology have the potential to provide prospective clinical studies of PART. Genetic studies so far have focused primarily on the frequency of the ApoE ε4 and MAPT H1/H1 haplotypes in PART and have reproducibly shown that the frequency of ApoE ε4 is less common than that seen in AD. Furthermore, the H1/H1 MAPT genotype is over-represented in PART, in contrast to AD [31••, 73].

One recent biomarker profile that has emerged in an attempt to originally capture preclinical AD is “suspected non-Alzheimer disease pathophysiology” (SNAP) [74, 75]. SNAP refers to cognitively normal or mildly impaired individuals who have normal β-amyloid biomarker levels (determined by CSF and PET ligand imaging) but abnormal neurodegeneration biomarkers (elevated levels of CSF p-tau, cerebral atrophy on MRI, and/or brain hypometabolism by ¹⁸FDG-PET). In the original study by Jack et al., 23% of over 70-year-old participants fulfilled the criteria of SNAP [75]. Several salient features of SNAP overlap with PART: (1) the frequency of SNAP increases with age [76], the ApoE ε4 haplotype is underrepresented in contrast to AD [75], and medial temporal atrophy is a feature of SNAP [74]. The clinical longitudinal data on SNAP appears mixed with some studies claiming that there are no significant cognitive differences between controls and individuals with SNAP, whereas other reports suggest steady cognitive decline in the setting of MCI in this group and worse cognitive performance than controls [77–80]. Further research is needed to better understand the clinical significance of SNAP and how well the biomarker profile of SNAP corresponds to the neuropathology of PART.

Advances in PET ligand imaging of tau also holds promise to better investigate PART antemortem. A recent PET imaging study utilized PET ligands for both tau and amyloid in a series of 301 individuals with either normal cognition or MCI who had also had recent cognitive testing [81]. Forty-five percent of participants were negative for amyloid but positive for tau (A−/T+), and these had less cognitive impairment than the A+/T+ individuals, who were considered to have early AD changes, but greater cognitive deficit than the A−/T− and A+/T− individuals. Importantly, their study of the A−/T+ group, which could be considered as the pathological equivalent of PART, showed more cognitive impairments than simply amnestic syndromes. Rather, they also demonstrated deficits in language and executive functioning which led the authors to speculate that the A−/T+ group represented individuals with early AD pathologic changes. More longitudinal studies with PET imaging could provide a better understanding of the natural history of these PET findings and address the next important question as to whether PART is simply an early manifestation of AD.

Does PART Represent a Neuropathologic Phenotype of Early AD?

On the one hand, with these clinicopathologic features already outlined in this review, there are clinicopathologic differences that distinguish classical AD and PART. The clinical course and cognitive deficits overall appear less severe than in age-and stage-matched AD patients. PART seems to affect an older demographic than classical AD, often in people over 80 years of age. The genetic and pathologic differences (absence of amyloid and limited Braak stage) are also in favor of separating PART from classical AD.

However, on the other hand, separation of PART from AD is not straightforward. One strong argument against PART being a new disease entity is that the pathology considerably overlaps with the neuropathologic changes of AD albeit without significant amyloid plaque burden. The prodromal period of AD is especially long, ranging up to several decades before symptom onset and amyloid deposition usually occurs many years after the first entangled neurons appear in cross-sectional post-mortem series [15, 82, 83]. As previously stated, NFT are a common phenomenon in autopsy brains and tauopathic changes have been demonstrated in childhood and early adulthood, particularly in the nucleus coeruleus, which has diffuse cortical and spinal connections, and within the transentorhinal cortex [14, 15]. These and other subcortical structures, such as the nucleus basalis of Meynert and the dorsal raphé nucleus, are considered by many experts to be the earliest sites of AD-related tauopathic changes and consistently show neuropathologic changes in AD [84, 85]. This overlap of selectively vulnerable nuclei between PART and AD in conjunction with the knowledge that the appearance of amyloid plaques occurs much later than do tangles challenges the notion that they are distinct disease processes but rather are the same disease process examined at different time points. Furthermore, the biochemistry and ultrastructure of the NFT in PART are similar to AD [25, 73, 86]. Likewise, immunohistochemistry against different tau isoforms has identified both 3R and 4R tau-labeled NFT in PART [31••]. Interestingly, a newly described phosphorylated tau binding protein, secernin-1, that binds to NFT with both 3R and 4R tau labels NFT in PART and AD, but not NFT associated with other tauopathies such as Pick’s disease (PiD) and progressive supranuclear palsy (PSP), where the NFT are composed of either 3R (PiD) or 4R (PSP) tau [86]. This concept of a continuum between PART and AD, such that PART represents early AD, has been proposed by Duyckaerts et al. [40••].

As mentioned, there are genetic distinctions between PART and classical AD with the H1/H1 MAPT genotype being over-represented in PART and the frequencies of ApoE4 gene carriers being less frequent than in AD [31••, 73, 87]. However, ApoE4 predisposes to beta-amyloidosis within the brain, and, therefore, this latter finding is expected given that selection of brains with PART selects against this haplotype. Moreover, while the H1/H1 MAPT genotype is over-represented in PART compared with AD in general, there is a similar frequency with limbic predominant AD [88, 89]. Some of the older studies included in the consensus PART paper may also have inadvertently included MAPT-associated FTLD-tau rather than PART.

Conclusions

At the very least, PART represents a subtype of AD that has a later onset of symptoms with a slower rate of disease progression. Further distinction of PART from AD as a distinct disease entity will rely on prospective biomarker and PET imaging studies. Also, additional work is needed in identifying a distinct biological mechanism that is separate to AD. A biochemical distinction between PART and AD associated NFT has yet to be accomplished. Outstanding questions related to the PART hypothesis include the following: (1) if tau acts in a prion-like fashion, why should NFT in PART be confined to Braak stage IV and not proteopathically spread further? [90] (2) Are there different tau strains in PART that contrast to other tauopathies? (3) What is the impact of amyloid plaques in the development of cortical NFT?

The morphologic changes in the brain with age are complex and variable between people. Aging is normally associated with the accumulation of certain proteins, such as neuromelanin in the substantia nigra [91]. In a similar manner, small amounts of tau aggregation in select nuclei in the brain could represent a usual phenomenon that steadily progresses with age. What is not so well understood is the rate and extent of this p-tau accumulation and how well clearance mechanisms are managing this. The cognitively intact 115-year-old woman referred to earlier in the review could be given the clinicopathologic diagnosis of asymptomatic, definite PART. Her rate of p-tau accumulation during her lifespan appears to be exceptionally slow for her age. Perhaps, addressing the rate of p-tau aggregation during the aging process and having an improved understanding of the pathophysiology of PART are keys to developing more personalized therapies for individuals who will develop cognitive impairment.