Cognitive and Neuropsychological Profiles in Alzheimer’s Disease and Primary Age-Related Tauopathy and the Influence of Comorbid Neuropathologies

INTRODUCTION

Alzheimer’s disease (AD) was first described microscopically in the early 20th century and is the most common cause of dementia worldwide [1, 2]. Alzheimer’s disease neuropathologic change (ADNC) is defined by the presence of hyperphosphorylated-tau (p-tau) neurofibrillary degeneration, a process which typically proceeds from medial temporal lobe structures into the neocortex in well-defined Braak stages [3], amyloid-β plaques (Aβ), which proceed from the neocortex to brainstem and cerebellum in Thal phases [4], and neuritic plaques (NP) in the neocortex [5]. These features, in particular Braak stage, correlate with cognitive status [5, 6], although there is growing evidence that some of the cognitive effects associated with this disorder are due, at least in part, to coexisting neuropathologic disorders, most commonly including limbic-predominant age-related TDP-43 encephalopathy neuropathologic change (LATE-NC), Lewy body disease (LBD), and cerebrovascular disease (CVD) [7 –19].

Primary age-related tauopathy (PART) is thought to be an Aβ-independent tauopathy that is primarily restricted to the medial temporal lobe, corresponding roughly from Braak stages I–IV in the absence of significant Aβ-deposition [20 –24]. “Definite” PART is currently defined as Braak stage I–IV in the complete absence of Aβ (Thal phase 0 and CERAD NP score “absent”), while “possible” PART is defined as Braak stage I–IV with minimal Aβ deposition (Thal phase 1-2 and/or CERAD NP score “sparse”) [20, 22]. Neurofibrillary degeneration in PART is thought to affect the CA2 hippocampal subregion early in the disease course, while in ADNC the entorhinal cortex and CA1 subregion are more severely affected with relative CA2 sparing [22 , 24–26]. Numerous studies have also shown that PART subjects differ from subjects with ADNC in terms of APOE ɛ2 and ɛ4 allele frequency [27 –31], MAPT haplotype [32], imaging characterization of brain atrophy patterns [33, 34], and clinical/cognitive features [27 , 35–39]. Clinically, PART patients have been shown to have relative preservation of attention, memory, language, and visuospatial function until later in the disease course, as well as a slower rate of cognitive decline after initial symptom onset compared to patients with autopsy-proven ADNC. We and others have demonstrated that cognitive function in subjects with PART is not significantly correlated with Braak stage, but rather the presence of hippocampal atrophy, white matter pathology, cerebrovascular disease, aging-related tau astrogliopathy (ARTAG), the presence and severity of LATE-NC, and the overall hippocampal tau burden [22 , 41], the latter of which is an observation confirming imaging studies suggesting an inverse correlation between medial temporal lobe tau levels, measured with positron emission tomography (PET), and cognitive performance [42]. Still, there remains debate as to whether PART is a distinct neuropathologic entity or belongs to the early stage of the ADNC spectrum [43 –45].

In this study, we leverage the National Alzheimer’s Coordinating Center database to compare demographics, genetics, neuropathologic features (including an array of comorbid disease states), cognitive features, and neuropsychological findings in autopsy-confirmed definite PART (n = 208), ADNC (n = 2,884), and “no tauopathy” (NT) (n = 178). We demonstrate significant differences in these profiles with and without neuropathologic comorbidities, suggesting that PART has a subtle but identifiable clinical correlate, with significant differences from ADNC.

METHODS

Case selection and exclusion criteria

For this study, we used the Uniform Data Set (UDS) and Neuropathology (NP) data set from the National Alzheimer’s Coordinating Center (NACC), established with funding from the National Institute on Aging (U01 AG016976). UDS and NP data were downloaded from NACC (https://naccdata.org/). Standardized UDS variable definitions [46] and NP variable definitions [47] from NACC were used, as described previously [48, 49]. A total of 7,709 unique NACC cases with last patient encounter within the final 24 months of life were identified [19]. In total, 3,803 cases were excluded for not having sufficient data to determine ADNC level. Of the remaining 3,906 cases, we then excluded 111 cases with progressive supranuclear palsy (NACCPROG), 72 cases with corticobasal degeneration (NACCCBD), 43 cases with Pick’s disease (NACCPICK), 4 cases with MAPT mutation (NPFTDT2), 184 cases with other, unspecified frontotemporal dementia (FTD)-Tau (NPFTDTAU), 70 cases with FTD-TDP (NPFTDTDP), 12 cases with chronic traumatic encephalopathy (CTE) (NPFTDT7), 87 cases with prion disease (NACCPRIO), 40 cases with amyotrophic lateral sclerosis/motor neuron disease (NPALSMND), 6 cases with Down syndrome (NACCDOWN), 5 cases with multiple system atrophy (NPPDXB), and 2 cases with trinucleotide repeat diseases (NPPDXD). Of note, a number of cases had multiple exclusionary criteria. Of the 3,270 cases remaining, 2,884 cases had intermediate or high level ADNC [5], 208 had definite PART [20, 22], and 178 had no identified tauopathy (Supplementary Figure 1). Demographic data on all individuals included in these three groups can be found in Table 1.

OPEN IN VIEWER

Fig. 1

Pie charts demonstrating the relative number of comorbid pathologies in cases with no tauopathy, PART, and ADNC.

OPEN IN VIEWER

Table 1

Demographic data in individuals with ADNC, PART, and no identified tauopathy

	No tauopathy (NT)	PART	ADNC	p (NT versus PART)	p (NT versus ADNC)	p (PART versus ADNC)
n	178	208	2884	–	–	–
Mean age (y)	78.6±0.5	82.0±0.8	80.2±0.2	=0.0006	=0.0496	=0.0203
Gender (M:F)	110:68	124:84	1572:1312	=0.6772	=0.0625	=0.1704
Race				=0.8391	=0.7393	=0.2707
Caucasian	93.7%	92.8%	94.3%
African American/Black	4.1%	5.3%	4.5%
Asian	1.6%	1.4%	0.8%
Hawaiian/Pacific Islander	0.4%	0.4%	0.1%
Education (y)	16.8±0.8	16.7±0.6	16.1±0.2	=0.9191	=0.3986	=0.4311
APOE Status
≥1 APOE ɛ2 allele	15.7%	24.5%	4.2%	=0.0425	<0.0001	<0.0001
≥1 APOE ɛ4 allele	12.4%	8.2%	51.3%	=0.1803	<0.0001	<0.0001

RESULTS

DISCUSSION

In recent years, there has been mounting evidence suggesting that primary age-related tauopathy, previously termed “tangle-only senile dementia” and “tangle-predominant senile dementia,” represents an entity distinct from AD, rather than simply a precursor to it, with studies demonstrating significant differences in the radiologic, neuropathologic, and genetic profiles between PART and ADNC [20 , 68]. While clinical symptoms and cognitive impairment in AD is thought to be most closely related to the topographic distribution of p-tau throughout the brain quantified by Braak staging, this does not appear to be the case with PART. Instead, the cognitive status in PART patients is determined primarily by overall p-tau-burden in the hippocampus, as well as the presence of comorbid neuropathologies, including white matter pathology, ARTAG, LATE-NC, and CVD [22 , 41]. Recent studies have shown that while PART does appear to have a deleterious effect on cognition, PART patients have a significantly slower rate of cognitive decline after becoming symptomatic comparted to patients with ADNC, and have relative sparing of a number cognitive domains, including semantic memory, language, and attention [31 , 38]. In this study, we used the NACC dataset to evaluate the cognitive profile of a cohort with definite PART, as compared to patients with no identified tauopathy (or other neurodegenerative pathologies) and patients with ADNC, and investigated the cognitive contributions of some of the more common neuropathologic comorbidities seen in PART.

When excluding LATE-NC, LBD, and CVD as comorbid pathologies, there were few significant differences in the cognitive profile between definite PART cases compared to NPI cases, including statistically equivalent global CDR, CDR sum of boxes, and MMSE tests (Table 4). The PART patients did perform significantly worse in measures of processing speed and executive function compared to the NPI cohort, and PART patients with Braak stage III-IV also had significantly worse performance in measures of language (Supplementary Table 2), while few meaningful differences were noted in patients with Braak stage I-II (Supplementary Table 3). Importantly, however, the majority of the cognitive features were significantly worse in ADNC than in PART (Table 4), allowing for a level of clinical discrimination between the two neuropathologic entities. It is interesting that the PART cohort performed worse than the NPI cohort in terms of WAIS CD, TMT-A, and TMT-B tests, which were previously shown to be progressively affected by Braak stage in PART patients [63]. These tests assess processing speed and executive function, but are also dependent on relatively intact visuospatial function, which may be affected in both PART and LBD, both of which show an early predilection for the CA2 hippocampal subregion [22 , 70]. The precise function of the CA2 subfield is unclear, but studies have shown that it has functions in social memory [71 –73], face-name pair encoding and retrieval [74], and visuospatial memory [39 , 76], suggesting that the particular cognitive and neuropsychological deficits found in this PART cohort may be due, at least in part, to the characteristic CA2 neurofibrillary degeneration pattern seen in these patients [22 , 25]. In addition, ADNC cases with low Braak stage (“early” AD) do not have selective deficits in the same cognitive domains as PART (Supplementary Tables 2, 3), suggesting that the particular differences in patterns of hippocampal pathology may result in different clinical phenotypes, particularly early in the disease course.

As demonstrated in numerous other studies [7 , 19], many autopsy-confirmed cases of ADNC have additional neurodegenerative findings that may affect cognition to various degrees. Our results demonstrate that LBD and various forms of CVD are more common in ADNC compared to PART and subjects without tauopathy, and more common in PART than patients without tauopathy, while the frequency of LATE-NC does not differ significantly between these groups (Table 2). These other neuropathologic findings further impair cognition in PART, ADNC, and NT cohorts in terms of CDR and MMSE (Table 3), and have variably deleterious effects on logical memory, attention, processing speed, language, and executive function (Table 5). Interestingly, while the presence of LATE-NC and CVD both cause cognitive impairment in subjects with PART, the presence of PART does not appear to cause additional cognitive impairment in subjects with LATE-NC and CVD, suggesting that these comorbidities may drive the more significant cognitive decline in PART patients.

As previous studies have also noted, the APOE profile differs between PART and ADNC [30 , 36]. In the current population, there are significantly more cases with at least one APOE ɛ2 allele and with two APOE ɛ2 alleles in the PART population and significantly fewer cases with one APOE ɛ2 and with two APOE ɛ2 alleles compared to the ADNC population. Given the role of APOE in the regulation of Aβ metabolism in the brain, the relatively reduced risk for Aβ accumulation with APOE ɛ2, and the increased risk for Aβ accumulation with APOE ɛ4 [77 –79], this differing genetic profile between PART and ADNC is not surprising. However, these findings do suggest that PART and ADNC may represent distinct disease processes, where the p-tau deposition in PART is driven by mechanisms unrelated to Aβ deposition, unlike in ADNC [21, 24]. In addition, the PART patients in this cohort have a significantly higher mean age compared to ADNC patients (and have been shown to represent the “oldest old” in other cohorts), suggesting that this tauopathy does not simply represent “pre-ADNC” which will eventually develop Aβ plaques and convert to ADNC [21, 45].

A significant limitation to this study is that the NACC dataset is not representative of the general population. Due to collection methods at nationwide Alzheimer’s Disease Research Centers, the cohorts are enriched for patients with more severe neuropathologic findings, more frequent and severe dementia, more APOE ɛ4 alleles, and more rare diseases compared to the general population. The population is also enriched for Caucasian patients with higher educational status than the general population and lacks a representative number control patients. Given the length of time NACC cases have been collected historically and the series of revisions to the NP and UDS variables, as well as the identification of new pathologic proteins, development of new antibodies, development of new classification systems, and recognition of new entities within this time period there are relevant pathologic variables which are unassessed for a subset of cases, particularly with LATE-NC, a diagnosis only formally codified in 2019 [51]. In addition, the pathologic variables assessed, particularly those related to AD pathology, assess primarily the overall distribution of pathology throughout the brain to a greater extent than the density or severity of pathology within a given region, which has been shown to be a particularly important predictor of cognitive status in PART [40, 42].

Our findings suggest clinical features which may represent the pure contribution of PART neuropathologic changes, and provide cognitive and neuropsychological differences between PART and AD. While PART cohorts have significantly better overall cognition than ADNC cohorts, processing speed and executive function appear to be selectively impaired in PART patients. Given the differing patterns of p-tau deposition in the hippocampus of subjects with PART and ADNC, as well as the more limited p-tau distribution and lack of Aβ, these clinical findings may help to provide functional insight into hippocampal subregions and other brain regions, as well as provide insight into mechanisms by which diffuse and neuritic Aβ plaques may contribute to cognitive impairment. Overall, these data help to further establish PART as an entity distinct from ADNC in terms of demographic, genetic, neuropathologic, and clinical features.

FUNDING

J.M.W. and T.E.R. are supported in part by National Institute on Aging (NIA) R21 AG078505 and Texas Alzheimer’s Research and Care Consortium (TARCC) grants 957581 and 957607. K.F. is supported in part by K01 AG070326.

CONFLICT OF INTEREST

Jamie M. Walker is an Editorial Board Member of this journal but was not involved in the peer-review process nor had access to any information regarding its peer-review. The other authors have no conflict of interest to report.

DATA AVAILABILITY

The data presented in this manuscript is derived from the National Alzheimer’s Coordinating Center (NACC) dataset, and is available upon request from https://naccdata.org/.