LATE-NC Stage 3: a diagnostic rubric to differentiate severe LATE-NC from FTLD-TDP

Abstract

A diagnostic rubric is required to distinguish between limbic-predominant age-related TDP-43 encephalopathy neuropathologic change (LATE-NC) and frontotemporal lobar degeneration with TDP-43 inclusions (FTLD-TDP). In LATE-NC Stage 3, TDP-43 proteinopathy is present in the middle frontal gyrus (MFG), thus posing a potential diagnostic challenge in differentiating these severe LATE-NC cases from FTLD-TDP. LATE-NC Stage 3 cases and other TDP-43 proteinopathies were analyzed from the University of Kentucky (total n = 514 with TDP-43 pathology assessed), The 90+ Study at the University of California Irvine (n = 458), and the Mayo Clinic (n = 5067) brain banks. Digital pathology was used to quantify pathology burden in a select subset of cases (n = 51), complemented by a previously-described manual counting method and expert neuropathologic examinations to evaluate qualitative features such as FTLD-TDP types and subtypes of neuronal cytoplasmic inclusions (NCIs). To evaluate clinical and genetic characteristics of LATE-NC Stage 3, data were analyzed from the National Alzheimer’s Coordinating Center (NACC) Neuropathology Data set and correlated with findings from the Alzheimer’s Disease Genetics Consortium (ADGC). When using TDP-43 proteinopathy quantification in the MFG as a diagnostic criterion, more than 90% of cases could be classified as either LATE-NC Stage 3 or FTLD-TDP. Diagnostically challenging scenarios included a subset of FTLD-TDP Type B cases with relatively mild MFG TDP-43 pathology and a novel non-LATE-NC, non-FTLD-TDP pathologic subtype with severe MFG TDP-43 pathology. Taking these potential pitfalls into account, a classification schema was developed that could correctly diagnose all included cases. There was no difference in the Alzheimer’s disease pathological load in LATE-NC Stages 2 versus 3. In genetic analyses, the GRN (rs5848) risk allele was preferentially associated with LATE-NC Stage 3, whereas TMEM106B and APOE risk-associated variants were not. In conclusion, LATE-NC Stage 3 could be differentiated reliably from FTLD-TDP and other TDP-43-opathies, based on a data-driven diagnostic rubric.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00401-025-02876-5.

Introduction

Neurological diseases with underlying TAR DNA-binding protein 43 (TDP-43) proteinopathy, collectively termed “TDP-43-opathies”, comprise a diverse spectrum of degenerative, trauma-related, and developmental disorders [9, 19]. The most prevalent subtype of TDP-43-opathy is limbic-predominant age-related TDP-43 encephalopathy (LATE), associated with LATE neuropathologic change (LATE-NC) [50]. Observed in over 30% of individuals older than 85 years of age at autopsy [49], LATE-NC manifests across a range of pathologic severity that correspond with varying levels of clinical impairment [44, 53, 84]. The present study investigated the pathologic features, clinical presentations, and genetic risk factors of the most severe subtype of LATE-NC.

A consensus group of experts recommended a pathology-based staging system for routine autopsy diagnosis of LATE-NC in 2019; the staging system was updated in 2023 [50, 52]. There are three recognized LATE-NC stages, based on the anatomic distribution of TDP-43 pathology: LATE-NC Stage 1 is characterized by any TDP-43 pathology in either the amygdala or hippocampal (when sparing the amygdala) regions; LATE-NC Stage 2 indicates that there is at least one TDP-43 immunoreactive neuronal cytoplasmic inclusion (NCI) in both the amygdala and the hippocampus regions; and, brains with LATE-NC Stage 3 have TDP-43 pathology in the amygdala region, hippocampal region, and middle frontal gyrus (MFG) [52]. This staging system was informed by prior work, notably by Dr. Keith Josephs and colleagues at the Mayo Clinic, who postulated a stereotypical progression of TDP-43 pathology in Alzheimer’s disease (AD) brains [27, 28].

Criteria for discriminating between the TDP-43-opathies is a work in progress. The first descriptions of TDP-43 proteinopathy were in amyotrophic lateral sclerosis (ALS) and frontotemporal lobar degeneration (FTLD) [58]. The subset of FTLD cases with confirmed TDP-43 pathology have been classified as frontotemporal lobar degeneration with TDP-43 pathology (FTLD-TDP), an umbrella term which is further subcategorized into FTLD-TDP Types A, B, C, D, and E based on pathologic and genetic features [57]. In addition to LATE-NC and FTLD-TDP, more than 20 other disorders have been shown to have TDP-43 pathology. Each disorder has characteristic TDP-43 pathologic features (histomorphology and anatomic distribution), as well as genetic risk factors, epidemiology, and clinical presentations [4, 9]. Although recognized histopathological hallmarks for these TDP-43-opathies have been described, distinguishing the diagnostic “boundary-zones” remains challenging. In particular, as stated by Alexandra Young et al. [86], there remains “considerable uncertainty in the early stages of the FTLD-TDP progression pattern and the late stages of the LATE-NC progression pattern.”

This uncertainty highlights the importance of systematic studies of LATE-NC Stage 3, which is a relatively common finding in dementia of the elderly. LATE-NC Stage 3 comprises ~ 10% of LATE-NC cases, and there is a ~ 4% lifetime risk of developing LATE-NC Stage 3 pathology in those who live beyond age 85 [49]. A key insight from Josephs et al. [27, 28] was that LATE-NC Stage 3—corresponding to their Stage 6—demonstrated pathology in the brainstem, diencephalon, basal ganglia, and cerebral neocortex, indicating a pattern of disease beyond limbic regions (Supplemental Fig. 1).

A critical anatomical region of interest for distinguishing LATE-NC from FTLD-TDP is the MFG. TDP-43 pathology is observed in the MFG in both LATE-NC Stage 3 and FTLD-TDP [52]. As a rule, the TDP-43 pathology in the MFG is much more severe in FTLD-TDP than in LATE-NC, but it was suggested that overlap may exist between severe LATE-NC and genetically-confirmed cases of FTLD-TDP [5, 65, 86]. Robinson et al. [86] proposed a method for distinguishing LATE-NC from FTLD-TDP based on TDP-43 lesion counts; however, the diagnostic reliability and pitfalls of using this method for differentiating LATE-NC from FTLD-TDP have not been verified.

In addition to the issue of overlap in the pathologic features between LATE-NC and FTLD-TDP, there are also questions about similarities and differences in the genetics and clinical manifestations between LATE-NC and FTLD-TDP. With regard to genetics, certain variants near the GRN and TMEM106B genes confer risk for both LATE-NC and FTLD-TDP. For example, the single nucleotide variant (SNV) in the GRN 3’ untranslated region, designated rs5848, has been reported as a risk factor for LATE-NC and in some (but not all) series of FTLD, as well as other conditions [7, 25, 26, 62, 71, 76, 77, 79]. Prior studies with comparisons between FTLD-TDP and LATE-NC also indicated group-level differences in epidemiological and clinical (e.g., cognitive and neuropsychiatric) parameters [65, 83, 86]. Despite progress in the field, more studies of LATE-NC are required to understand the disease-specific features of this disorder.

In the present study, we investigated pathologic, clinical, and genetic characteristics of LATE-NC Stage 3. We quantified and analyzed TDP-43 pathology in both LATE-NC Stage 3 and FTLD-TDP brains. Clinical and genetic data for LATE-NC Stage 3 were drawn from the National Alzheimer’s Coordinating Center (NACC) and Alzheimer’s Disease Genetic Consortium (ADGC) datasets. We aimed to test the possibility of developing a diagnostic rubric for differentiating severe LATE-NC from FTLD-TDP.

Methods

Study participants

For detailed pathologic investigations, human brain tissue samples were analyzed from research participants who had gone through clinical and pathological evaluation from the University of Kentucky Alzheimer’s Disease Research Center (UK-ADRC) brain bank (UK-BB), Mayo Clinic brain bank (Mayo-BB) and University of California-Irvine The 90+ Study brain bank (UCI-BB). Information related to the cohorts, and an overview of neuropathologic methods, have been published elsewhere [36, 67, 69, 74]. All studies were performed with IRB approval and research using anonymized autopsy samples was considered exempt from human subject research designation. Each case was given a clinical diagnosis, followed by a neuropathologic classification based on the clinicopathological consensus. Samples from the UK-BB were selected based on the presence of the most severe level of LATE-NC Stage 3 (n = 9). Also assessed from UK-BB were autopsy-confirmed cases of FTLD (n = 3), representing the only cases of FTLD-TDP types A or B from the UK-BB. Similarly, all LATE-NC Stage 3 cases from the UCI-BB were used for the current study (n = 13). Of the 5,067 Mayo-BB cases screened using TDP-43 immunohistochemistry, 1,690 were positive for TDP-43 pathology. The 27 included Mayo-BB cases were a convenience sample that spanned a variety of TDP-43-opathies, including 10 FTLD-TDP type A (one GRN variant carrier), 10 FTLD-TDP type B (6/10 with MND and one with C9ORF72 repeat expansion), 5 LATE-NC/FTLD-TDP borderline cases, and 2 CBD-TDP cases. The criteria for selecting the 27 cases from the Mayo-BB were that all cases were Braak neurofibrillary tangle (NFT) Stage IV or lower, Thal Aβ Phase of 3 or lower, and all cases lacked additional clinically significant pathologies (excluding the two CBD-TDP cases). After the unstained slides were received at the UK-ADRC, the UK-ADRC raters who evaluated them were blinded initially to all clinical and pathological information.

Immunohistochemistry

The focal-point of the study was specific pathological features in formalin-fixed, paraffin-embedded (FFPE) tissue from the MFG (Brodmann area [BA] 9). Sections were cut from FFPE tissue blocks on a microtome at a thickness of 10 μm and mounted on glass microscope slides. Immunohistochemistry was performed as described previously in detail [48]. Briefly, phosphorylated TDP-43 immunoreactivity was visualized using the ID3 antibody (1:500) using a chromogenic substrate (Nova Red, Vector Laboratories, Cat #SK-4800) [56]. For evaluation of Alzheimer’s disease neuropathologic changes (ADNC) using digital pathology, the methods used were as described before in detail [55, 59].

Analysis of TDP-43 pathology

After immunohistochemical staining, sections were scanned and digitized with a Zeiss Axio Scan Z.1 slide scanner. Images were captured at 20× magnification, with a Z-stack of 12 images spaced 1 μm apart. Zen 2.6 Blue Edition software was used to merge the images into a two-dimensional image using the contrast method. All cases were initially scored by a blinded rater (RSK) using the method that was described in detail by Robinson et al. [65]. Briefly, the scoring method employs a diagnostic threshold for distinguishing LATE-NC from FTLD-TDP based on light microscopic evaluation by a blinded neuropathologic rater. Any case with >15 TDP-43 proteinopathic structures (TDP-43 immunopositive neurites or NCIs) detected in at least 3 high-power (40× magnification) microscope fields of MFG was considered to mostly likely be FTLD-TDP. When MFG TDP-43 pathology did not meet that threshold the case was considered LATE-NC if other exclusionary criteria did not apply (see below).

For digitally-generated counts for TDP-43 pathology, the Object Colocalization tool was used in HALO software on the original brightfield images. Results presented are based on the number of TDP-43+ structures present per mm² in each section, using the following parameters that were applied to all sections: stain intensity, signal to noise ratio, and relative cell size-including processes. To ensure that all TDP-43+ structures were counted, the size threshold was set between 1 µm and 1000 µm. Following digital detection, structures deemed not diagnostically relevant were excluded post-hoc based on average optical density and size. Exclusion criteria in HALO software were applied: an inner diameter ≥ 1 µm, an outer area (µm²) ≥ 6 µm, and an average optical density ≥ 0.3. After setting detection parameters, detection boxes were placed manually in specific layers of MFG cortical gray matter, i.e., superficial layer, middle layer and deep layer (n = 5 boxes per layer, 15 boxes per section).

In terms of eventual clinicopathological classification, the diagnoses for UK-BB and UCI-BB cases were known prior to analyses, and all except three were assessed to be LATE-NC Stage 3. The Mayo-BB cases were read (and pathology quantified at the UK-ADRC) blind to clinical or pathologic data. Post hoc classifications were made that included clinical information. For cases over the age of 75 years at death, LATE was diagnosed if there was documented memory impairment as a primary complaint, and no significant behavioral, language or movement disorders (e.g., FTD, PPA, ALS, MSA, CTE, multiple sclerosis, prion disease, Huntington disease or other triplet repeat disorder) [82].

National Alzheimer’s Coordinating Center (NACC) data

Brain autopsies were performed at each ADRC that contributed cases. The included data from the NACC Uniform Data Set (UDS) [2] and Neuropathology (NP) [3] Data Set data were entered before June 2024. The NP data were measured via the NACC NP v10-11 forms and included data from 37 different National Institute on Aging-funded ADRCs. We excluded data from participants who were younger than 70 years of age at death in the NACC NP Data Set. We also excluded participants who were diagnosed with any of 19 rare brain diseases at autopsy (Supplemental Table 1). The LATE-NC stages were determined by brain region distribution of TDP-43-immunoreactive inclusions as described previously [33].

Cognitive data were drawn from the NACC UDS test battery [2], including Mini Mental State Examination (MMSE) [18], Animal [42] and Boston Naming Tests [29], Wechsler Memory Scale-Revised (WMS-R) Logical Memory immediate and delayed recall [81], and Digit Span Forward and Backward trials [80]. Since the Montreal Cognitive Assessment (MoCA) [46], Multilingual Naming Test (MINT) [23], Number Span Test: Forward and Backward, and Craft Story 21 Recall—immediate and delayed [11] were introduced in the NACC UDS version 3 from March 2015 (neuropsychological battery—form C2) instead of MMSE, Boston Naming Test, Digit Span Forward and Backward trials, and WMS-R Logical Memory—immediate and delayed (neuropsychological battery), respectively, we transformed the new battery scores into equivalent old battery scores based on Monsell and colleagues’ crosswalk study [40]. We also included the Clinical Dementia Rating Scale (CDR) Sum of Boxes ratings with the cognitive test measures.

Neuropsychiatric symptoms (NPS) were measured in the UDS using the Neuropsychiatric Inventory (NPI-Q) [34, 48]. Study co-participants (defined as someone who knew the participant well, usually a caregiver for persons with dementia) were asked if the participants had the following specific NPS in the past month prior to the study visit: delusions, hallucinations, agitation or aggression, depression/dysphoria, anxiety, elation/euphoria, apathy/indifference, disinhibition, irritability/lability, motor disturbance, nighttime behaviors, and appetite and eating problems.

Alzheimer’s Disease Genetics Consortium (ADGC) genetic data

Genetic data were obtained from Alzheimer’s Disease Genetics Consortium (ADGC), and were linked to clinical and neuropathological outcome data on each study participant from the NACC data set as described earlier [32]. The genotype data were imputed using the TOPMed Imputation Server (https://imputation.biodatacatalyst.nhlbi.nih.gov/) based on Genome Reference Consortium Human Build 38 (GRCh38) [12]. We examined four single nucleotide polymorphisms (SNPs): rs13237518 (a proxy SNP for rs1990622) in TMEM106B, rs5848 in GRN, and rs7412 and rs429358 in APOE.

Data analysis

Data were presented as mean ± standard error of the mean (SEM) unless otherwise specified. Heatmaps were generated using JMP Pro Software version 17.0 and were generated using Ward hierarchical clustering based on pathological burden (pathologic TDP-43 structures per mm²), standardized by layer, and subsequently visualized based on pathological phenotypes present as well. We conducted multivariable linear and logistic regression analyses for the cognitive test scores and NPS, adjusted for age at last visit, sex, years in education, ADNC, and Lewy bodies. To examine the genetic association of the four SNPs with LATE-NC Stages 2 and 3, we applied multivariable logistic regression models adjusted for age at death, sex, ADNC, Lewy bodies, and the first three principal components assuming additive mode of inheritance as described earlier [32]. Data and statistical analyses were conducted using JMP Pro Software version 17.0 (SAS Institute, Cary, NC, USA), GraphPad Prism software version 10, or R version 4.4.1. Statistical significance was preset at an α level of 0.05.

Results

Study workflow and participants

The overall workflow for the present study is shown in Fig. 1. Human brain tissue samples were analyzed from three different brain banks: UK-BB, UCI-BB, and Mayo-BB. All of the analyses, unless specified otherwise, focused on phospho-TDP-43 immunohistochemistry (IHC) assessed in the MFG. Summary information for each of the selected cases is shown in Table 1. More specifically, after developing the TDP-43 pathologic quantification method, we first studied cases from the UK-BB. From this cohort we analyzed the most severe (high-TDP-43 histopathology in the MFG) cases of LATE-NC Stage 3 (n = 9) and autopsy-confirmed cases of FTLD-TDP Types A or B (n = 3). Notably, there were 10 other cases of LATE-NC Stage 3 from the UK-BB (total of n = 19 LATE-NC Stage 3 cases) that had negligible amounts of MFG TDP-43 proteinopathy. These 10 cases were not included subsequently in the present study. (Because of the low MFG TDP-43 pathology levels in these cases, in combination with their clinical histories of amnestic dementia, they were identified as LATE-NC.) Next, autopsy-confirmed cases, representing all of the LATE-NC Stage 3 cases from the UCI-BB (n = 13), were selected. One of these 13 cases had tissue that was technically incompatible with IHC analyses and so was excluded, leaving n = 12 cases from UCI-BB. Finally, we tested the MFG TDP-43 proteinopathy in a mixed cohort of TDP-43-opathies (n = 27) from the Mayo-BB, and the outcomes were used collectively to assess whether or not a coherent diagnostic rubric could be generated to differentiate LATE-NC Stage 3 from other TDP-43 pathologic entities.

Fig. 1.

Study workflow illustrating the approach used to compare and contrast LATE-NC Stage 3 with other TDP-43-opathies. Cases for neuropathological studies (TDP-43 immunohistochemical staining of middle frontal gyrus) were sourced from three brain banks: the U.Kentucky (UK-BB), The 90+ U.California Irvine(UCI-BB), and Mayo Clinic (Mayo-BB) brain banks

Table 1.

Cases from University of Kentucky ADRC brain bank (UK-BB), The 90+ Study and University of California Irvine ADRC brain bank (UCI-BB), and Mayo Clinic brain bank (Mayo-BB) for assessment of middle frontal gyrus (MFG) TDP-43 pathology

AD Probable Alzheimer’s disease ALS Amyotrophic lateral sclerosis, ARTAG Age-related tau astrogliopathy, CBS/D Corticobasal syndrome/degeneration, CIND cognitive impairment no dementia (equivalent to MCI), DLB Dementia with Lewy bodies, FTD Frontotemporal degeneration, FTD-MND Frontotemporal lobar degeneration with motoneuron disease, HS Hippocampal sclerosis, LATE Limbic-predominant age-related TDP-43 encephalopathy, Mayo-BB Mayo Clinic ADRC brain bank, MFG Middle frontal gyrus, NPH Normal pressure hydrocephalus, PD Parkinson disease, PPA Primary progressive aphasia, PNFA Progressive non-fluent aphasia, PSP Progressive supranuclear palsy, UCI-BB The 90+ Study and University of California Irvine ADRC brain bank, UDS Uniform Data Set, UK-BB University of Kentucky ADRC brain bank, VD Vascular dementia

Representative depictions of the MFG TDP-43 pathologies are shown for FTLD-TDP Type A, FTLD-TDP B, and both high- and low-severity LATE-NC (Fig. 2). Note that the density of MFG TDP-43 proteinopathy detected for FTLD-TDP Type A was higher than in FTLD-TDP Type B. Also, qualitatively distinctive histomorphologic TDP-43-immunopositive structures were observed (representative photomicrographs of these microscopic lesions are depicted in Supplemental Fig. 2). Quantification of TDP-43+ structures in the UK-BB and UCI-BB cohorts is shown in Fig. 3. Corresponding H&E stained sections are shown in Supplemental Fig. 3.

Fig. 2.

Distribution of detected TDP-43-immunoreactive structures in the middle frontal gyrus (MFG) in FTLD-TDP and LATE-NC. Shown are representative examples of the distribution of TDP-43 immunoreactive structures (orange dots) in FTLD Type-A, FTLD Type-B, and LATE-NC (High- and Low-Pathology), across cortical layers (I–VI) and the superficial white matter (WM) of MFG. FTLD Type-A displays a relatively dense accumulation of pTDP-43 immunoreactive structures. In comparison, FTLD Type-B cases tended to demonstrate a sparser distribution. Even in LATE-NC with relatively high pathology, TDP-43 proteinopathy had lower overall burden compared to FTLD cases

Fig. 3.

Quantification of middle frontal gyrus (MFG) TDP-43-immunoreactive structures in University of Kentucky (UK) and The 90+ University of California Irvine (UCI) brain banks. Counts of of TDP-43+ structures were evaluated across superficial (a), middle (b), and deep cortical layers (c) of MFG in cases of LATE-NC (UK and UCI cohorts) and FTLD-TDP (UK cohort). Data are plotted on a logarithmic scale (y axis indicates TDP-43+ structures per mm²). A threshold was noted for distinguishing between LATE-NC and FTLD-TDP, represented by the dotted line. The individual data-point marked by a blue diamond (a–c) represents a case with potential limbic neuroastroglial tauopathy (LNT), with severe aging-related tau astrogliopathy (ARTAG) in the hippocampus (panel d and see Supplemental Fig. 4) whereas the boxed data-point (a–c) represents a case with many gNCIs. The high-magnification photomicrograph in panel (e) depicts a gNCI TDP-43 + structure. Scale bars = 15 μm in (d), 10 μm in (e)

With counts of TDP-43 pathology being generated at UK-ADRC, while blinded to any clinical or pathological information, we next evaluated the various TDP-43-opathy cases from Mayo-BB. Quantification of TDP-43 pathologic subtypes in all included cases was stratified by cerebral cortical layers, and are shown in heatmaps following hierarchical clustering analysis that was based on pathological load and patterns. Figure 4 includes quantitative TDP-43 pathology counts (blue squares) and evaluations of qualitative TDP-43 pathologic features (e.g., NCIs; red squares), as well as the clinical-pathological features (e.g., motoneuron disease) of each case.

Fig. 4.

Hierarchical clustering of TDP-43 pathology in MFG cortical layers for all included cases. In the leftward boxed column, cases that met the criteria for FTLD-TDP based on the presence of > 15 TDP-43 immunoreactive lesions per high-powered field of view (40x) were designated with a gray box, cases that TDP-43 severity compatible with LATE-NC with a white box. Clustering was based on levels of TDP-43 pathological burden (blue) and distribution of specific TDP-43+ pathology types across superficial, middle, and deep layers (red). Color intensity in the red heatmaps indicates the relative proportion of each pathology type, with deeper red denoting higher pathologic burden. Cases that were identified as FTLD Type B or C, FTLD-MND, or CBD/FTLD are so marked. Cases marked by a blue diamond represent potential limbic neuroastroglial tauopathy (LNT). Note that all LATE-NC cases could be discriminated based on a combination of clinical and pathologic criteria

Following digitally-obtained quantitative analyses of the severity of TDP-43 pathology in the MFG, an apparent upper threshold for LATE-NC and a lower limit for FTLD-TDP were detected (Figs. 3, 5, and 6). There appeared to be differentiation between LATE-NC and FTLD-TDP at the TDP-43 lesion density of ~ 100 structures per mm² in the superficial cortex of the MFG. In other words, most cases from all 3 brain banks could be reliably designated as either LATE-NC Stage 3 or some other condition based on TDP-43 immunohistochemical quantification in the superficial layers of the MFG.

Fig. 5.

Distribution of TDP-43 pathology in different layers of cerebral cortical (MFG) gray matter—Mayo Clinic brain bank cases—stratified by TDP-43 pathologic subtypes. The densities of TDP-43+ pathological structures are shown across MFG cortical layers—superficial, middle, and deep cortical layers. Data points represent individual cases, color-coded by autopsy-confirmed subtype: FTLD-TDP Type A (gray), FTLD-TDP Type B (red), FTLD-TDP Type C (yellow), and probable LATE-NC (cyan). The proposed cutoff between LATE-NC and FTLD is marked by a dashed line at a density of > 100 TDP-43+ structures per mm.². Cases labeled as “probable LATE-NC” represent individuals that met the diagnostic criteria for LATE-NC based on clinicopathological history. The single high-pathology case marked by blue-colored diamonds indicates a case with potential limbic neuroastroglial tauopathy (LNT)

Fig. 6.

Distribution of TDP-43 pathology in different layers of cerebral cortical (MFG) gray matter—Mayo Clinic brain bank cases—stratified primarily by clinical-pathological subtypes. Here, cases are color-coded based primarily on predicted clinical classification: FTD/FTLD (gray), FTD-MND (yellow), CBD/FTD (red), and probable LATE-NC (cyan). The single case highlighted with diamond shapes represents a possible case of limbic neuroastroglial tauopathy (LNT)

Potential challenging scenarios when differentiating LATE-NC Stage 3 from FTLD-TDP

In a minority of cases, there was initial diagnostic ambiguity such that LATE-NC and FTLD-TDP cases seemed to overlap in terms of MFG TDP-43 proteinopathic burden. More specifically, some individuals with a diagnosis of FTD and/or motoneuron disease by clinical criteria had MFG TDP-43 proteinopathy that was quantitatively similar to the levels seen in cases with LATE-NC. Moreover, there were 2 cases that met clinical criteria for LATE, but had relatively high levels of TDP-43 proteinopathy in MFG. We evaluated these outliers on a case-by-case basis. In differentiating LATE-NC Stage 3 from other TDP-43 pathologies, three potential diagnostic challenges were identified:

• FTLD-TDP Type B with granular-type TDP-43 neuronal cytoplasmic inclusions (gNCIs): Cases with gNCIs tended to be counted as low-pathology, with quantification readouts that sometimes overlapped with LATE-NC. On the other hand, the diagnostic histomorphology of gNCIs could be readily recognized via light microscopy by a neuropathologist.

• FTLD-MND: These cases had heterogeneous MFG TDP-43 proteinopathy but there was a consistent clinical-neuropathological correlation—a history of motoneuron-referent symptomatology with autopsy confirmation of TDP-43 pathology in motoneurons.

• Presumed limbic-predominant neuronal inclusion body 4R tauopathy (LNT) cases: This is a relatively newly characterized entity [1, 37–39, 68] with a high level of MFG TDP-43 pathology that could overlap with MFG TDP-43 pathologic burden in FTLD-TDP, but in patients that fit better clinically with AD or LATE. These cases are distinguished by a very heavy amount of MTL tau age-related tau astroglial pathology.

Each of these scenarios are described in greater detail, below.

Some cases of FTLD-TDP type B had lower densities of TDP-43 immunoreactive structures (when quantified via digital pathology methods) across cerebral cortical layers compared to FTLD-TDP type A. Some even had similar TDP-43 proteinopathy levels as LATE-NC (Fig. 3, 4, and 5). Evaluating these “overlap” cases more closely, qualitative diagnostic assessments revealed a subset of FTLD-Type B cases with gNCIs. These could be distinguished from other diffuse NCI (dNCI) [57], based on the distinctive “powdery” perinuclear ring of TDP-43 immunoreactivity (Fig. 3e). Importantly, this phenotype was not seen in any LATE-NC Stage 3 case (Fig. 4). These findings suggest that gNCIs (a qualitative observation that remains to be verified by, for example, machine learning approaches) provides another criterion for distinguishing FTLD-TDP Type B from LATE-NC.

Another subset of the non-LATE-NC cases with overlapping levels of pathological MFG TDP-43 pathologic burden, in comparison to LATE-NC, were patients with clinical-pathologic diagnoses of FTLD-MND (Fig. 6). Notably, FTLD-MND cases can have either relatively low or high levels of TDP-43 burden in the MFG (see Fig. 6.) However, all FTLD-MND cases were correctly diagnosed and classified after using a TDP-43 IHC stain on motoneurons in the spinal cord or cerebral cortex. Analogously, some corticobasal degeneration (CBD) cases had similar levels of TDP-43 burden with LATE-NC (Fig. 6). As such, cases with a clinical history compatible with FTD, CBD or motoneuron disease should be identified a priori and, in those cases, a primary diagnosis other than LATE-NC should be considered.

There also were 2 included cases that had relatively severe MFG TDP-43 pathology despite a clinical history that was compatible with LATE (i.e. amnestic dementia). Intriguingly, both of these cases had features that define a distinct, recently-characterized pathologic entity, other than LATE-NC [1, 37–39, 68]. The terminology for this condition is limbic-predominant neuronal inclusion body 4R tauopathy (LNT) [70]. Both of these cases had extremely robust hippocampal tau astrogliopathy. Shown in Fig. 3, and, in greater detail in Supplemental Fig. 4, are clinical, genetic, and pathological data from a 91-year old with homozygous risk alleles for TMEM106B, a case that was previously (in retrospect, mis-) diagnosed at the UK-ADRC as LATE-NC Stage 3.

Overall comparisons between TDP-43-opathy subtypes using TDP-43 pathology counts and integrating clinical information

The above criteria could be cross-referenced with the “hand-counting” of TDP-43 pathology as recommended by Robinson et al. [65]. The results of those counts are shown on the leftward (gray or white boxes) column of Fig. 4, along with clinical and other pathologic data. Using these methods for diagnostic categorization, each included case in the present series was correctly classified as either LATE-NC or some non-LATE-NC condition (e.g., FTLD-MND, FTLD-TDP type B, or LNT).

Neuropathological profiles of Aβ and tau pathologies (ADNC) across LATE-NC Stages

We sought to determine if LATE-NC Stage 3 cases had more or less severe ADNC in comparison to LATE-NC Stage 2. The relationship between LATE-NC stages and concomitant ADNC was analyzed in 417 UK-ADRC cases, focusing on Braak NFT stages and digitally-quantified Aβ amyloid and tau (pTau) pathologic burdens. There was a positive correlation between LATE-NC and higher Braak NFT stages (Fig. 7a). Quantitative regional analyses of Aβ and pTau burdens across hippocampal (CA1), frontal (BA9), and parietal (BA40) cerebral cortical regions (Fig. 7b–f) suggested that while Aβ amyloid and tau pathologic burdens were increased in advanced LATE-NC stages in the frontal and parietal cortices, Aβ and pTau burdens remain relatively stable across LATE-NC Stages 2 and 3. These results indicate that, while there is apparently a positive association between LATE-NC and ADNC (LATE-NC often co-occurs in cases with high levels of ADNC), the severity of ADNC was not changed in LATE-NC Stage 2 versus Stage 3 (Fig. 7f).

Fig. 7.

Stratifying Alzheimer’s disease neuropathologic changes (ADNC) by LATE-NC stages. Panel (a) shows neuropathological characteristics of UK-ADRC cases (N = 417) stratified by LATE-NC stages and Braak neurofibrillary tangle (NFT) stages. Percentages of cases in LATE-NC stages 2 and 3 are provided for each Braak NFT stage. ADNC was also assessed with digital methods to quantify immunohistochemical (IHC) staining in specific brain regions, as previously described. Representative images of PHF-1 tau (panel b) IHC and its digital markup (panel c), and β-amyloid plaques (panel d) IHC and their digital markup (panel e) are shown. A quantitative comparison of Abeta and phospho-tau (pTau) burdens across LATE-NC stages in the hippocampus (CA1), frontal cortex (BA9), and inferior parietal lobule (BA40) are shown in bar charts (panel f). There was generalized increases in ADNC pathology in advanced LATE-NC stages. However, there was not ADNC severity differences detected between LATE-NC Stages 2 and 3. Error bars = StDev

Clinical features comparing LATE- NC Stage 2 and LATE-NC Stage 3

Since this study was focused on features of LATE-NC Stage 3, we wanted to compare between LATE-NC Stage 2 and Stage 3 in terms of clinical manifestations. These were evaluated in a convenience sample from the NACC NP Data Set comprising n = 571 participants (n = 459 LATE-NC Stage 2; n = 112 LATE-NC Stage 3) with summary information on these individuals shown in Table 2. Overall mean age of death was 86.0 years ± 8.0 years, and the sample was 55% females. This cohort was biased toward those with dementia (88% had documented dementia at last clinical evaluation). Cognitive function tended to be lower in LATE-NC Stage 3 than Stage 2 in this sample, but these trends were not statistically significant (Table 3). For example, the final MMSE scores were on average 1.4 points lower in LATE-NC Stage 3 compared to Stage 2 (P = 0.27). Likewise, in terms of neuropsychiatric symptoms (Table 4), the only symptom significantly different in between LATE-NC Stages 2 and Stage 3 was hallucinations (P < 0.05), and this finding was marginal considering that the statistical test results were not corrected for having performed multiple comparisons. Notably, neither disinhibition nor primary progressive aphasia (both associated with the clinical syndrome of FTD) were different between individuals with LATE-NC Stages 2 and 3.

Table 2.

Characteristics by stages in subjects of National Alzheimer’s Coordinating Center (NACC) for clinical comparisons between LATE-NC Stages 2 and 3

Characteristics	Overall (n = 571)	Stage 2 (n = 459)	Stage 3 (n = 112)
Age at death, mean ± SD	86.0 ± 8.0	85.7 ± 8.1	87.3 ± 7.5
Years in education, mean ± SD	15.9 ± 2.9	16.0 ± 2.8	15.5 ± 2.9
Sex, n (%)
Male	255 (44.7)	203 (44.2)	52 (46.4)
Female	316 (55.3)	256 (55.8)	60 (53.6)
Cognitive status at UDS visit, n (%)
Normal	26 (4.6)	22 (4.8)	4 (3.6)
Impaired-not-MCI	7 (1.2)	7 (1.5)	0 (0)
MCI	36 (6.3)	32 (7.0)	4 (3.6)
Dementia	502 (87.9)	398 (86.7)	104 (92.9)
APOE genotypea, n (%)
−/−	213 (41.2)	166 (39.8)	47 (47.0)
−/ε4	250 (48.4)	203 (48.7)	47 (47.0)
ε4/ε4	54 (10.4)	48 (11.5)	6 (6.0)

UDS Uniform Data Set, MCI mild cognitive impairment, SD standard deviation

^an = 517 had APOE genotype data

Table 3.

Associations with cognitive functions at last visit in LATE-NC Stage 3 compared with LATE-NC Stage 2

Cognitive test	$\widehat{\beta }$a	SE	P-value
CDR sum of boxes	1.28	0.67	0.059
Mini mental state examination	− 1.47	1.30	0.26
Animal naming test	− 0.71	1.23	0.56
Boston naming test	− 0.99	1.84	0.59
Logical memory—immediate	− 0.047	0.96	0.96
Logical memory—delayed	0.056	0.99	0.96
Digit span forward trials	− 0.82	0.51	0.11
Digit span forward trials	− 0.044	0.48	0.93

LATE-NC Limbic-predominant age-related TDP-43 encephalopathy neuropathologic change; CDR Clinical Dementia Rating Scale; SE standard error

^aAdjusted for age at last visit, sex, years in education, Alzheimer’s disease neuropathological changes (ADNC), and Lewy bodies

Table 4.

Associations with neuropsychiatric symptoms at last visit in LATE-NC Stage 3 compared with LATE-NC Stage 2

Symptom	OR (95% CI)a	P-value
Delusions	1.52 (0.81–2.88)	0.19
Hallucinations	2.06 (1.04–4.07)	0.038
Agitation or aggression	1.72 (0.99–2.97)	0.052
Depression or dysphoria	1.03 (0.59–1.81)	0.91
Anxiety	1.21 (0.69–2.13)	0.51
Elation or euphoria	2.51 (0.83–7.64)	0.10
Apathy or indifference	0.78 (0.45–1.33)	0.36
Disinhibition	1.56 (0.86–2.81)	0.14
Irritability or lability	1.03 (0.59–1.81)	0.91
Motor disturbance	1.22 (0.66–2.25)	0.52
Nighttime behaviors	1.44 (0.83–2.49)	0.20
Appetite and eating problems	0.72 (0.39–1.34)	0.31
Indicate whether the subject currently is meaningfully impaired, relative to previously attained abilities, in language	1.07 (0.58–1.98)	0.83
Primary progressive aphasia (PPA) with cognitive impairment	0.51 (0.05–4.84)	0.56

LATE-NC Limbic-predominant age-related TDP-43 encephalopathy neuropathologic change, OR odd ratio, CI confidence interval

^aAdjusted for age at last visit, sex, years in education, Alzheimer’s disease neuropathological changes (ADNC), and Lewy bodies

Genetic variants associated with LATE-NC stage 3

Genetic analyses of known polymorphisms in TMEM106B (rs13237518 A allele [a proxy for rs1990622 allele status]), GRN (rs5848 T allele), and APOE (rs7412 T allele and rs429358 C allele, which together represent APOE ε4 status) were performed to assess whether they were associated with the progression to LATE-NC Stage 3, focusing on comparing cases with LATE-NC Stage 2 and Stage 3 (Table 5, Fig. 8 and Supplemental Fig. 5). The TMEM106B rs13237518 A allele demonstrated a variable relationship with LATE-NC (the more common allele is the risk-associated allele, in this cohort). While heterozygous carrier (1 A allele) frequency was higher in Stage 3 (23.4%) compared to non-carriers (17.8%), homozygous carrier frequency (2 A alleles) was lower in Stage 3 (16.0%). These findings suggest a complex relationship between TMEM106B genotype and pathological progression. By contrast, a more straightforward-seeming dose-dependent association was observed between the GRN rs5848 T allele and LATE-NC Stage 3 (Fig. 8). Non-carriers had a lower proportion of LATE-NC Stage 3 pathology (13.1%) compared to carriers of one (22.8%) or two (34.0%) T alleles. The statistical analyses of these results are depicted in Table 5. These findings suggest that carrying a GRN rs5848 T allele may accelerate pathological progression from LATE-NC Stage 2 to Stage 3. By contrast, APOE polymorphisms (rs7412 T and rs429358 C) showed allelic dosage distributions that were similar between LATE-NC Stage 2 and Stage 3. As such, our data do not indicate that APOE ε4 status was associated with progression to LATE-NC Stage 3 (Fig. 8 and Supplemental Fig. 5).

Table 5.

Associations of SNPs in LATE-NC Stage 3 vs. LATE-NC Stage 2

Chr	SNPs	Gene	Effect allele	OR (95% CI)a	P-value
7	rs13237518	TMEM106B	A	1.01 (0.95–1.07)	0.81
17	rs5848	GRN	T	1.11 (1.04–1.18)	0.00094
19	rs429358	APOE	C	0.99 (0.92–1.06)	0.70
19	rs7412	APOE	T	0.94 (0.81–1.10)	0.45

LATE-NC Limbic-predominant age-related TDP-43 encephalopathy neuropathologic change, Chr chromosome, SNP Single nucleotide polymorphism, OR odd ratio, CI confidence interval

^aAdjusted for age at death, sex, Alzheimer’s disease neuropathological changes (ADNC), Lewy bodies, and the first three principal components

Fig. 8.

Associations between genetic variants and LATE-NC Stage 3 (versus Stage 2). Stacked bar charts illustrate the relationship between genetic polymorphisms in TMEM106B (rs13237518 A allele; panel a), GRN (rs5848 T allele; panel b), and APOE (rs7412 T allele and rs429358 C allele; panels c, d) when comparing LATE-NC Stage 2 vs. Stage 3. Each bar plot displays the percentage distribution of LATE-NC cases across genetic allele dosages (0, 1, or 2 minor alleles) for each variant. Stage 2 (pink) and stage 3 (blue) cases are presented as proportions within each allele category. These data suggest a potential genetic influence of GRN variants (particularly in people with homozygous risk alleles) on the progression to advanced LATE-NC pathology, while TMEM106B and APOE variants appear less impactful in terms of the difference between LATE-NC Stages 2 and 3 in this cohort. For additional data on genetic associations relevant to LATE-NC Stages (0-3) see Supplemental Fig. 5

Discussion

The present study focused on pathological, clinical, and genetic features of LATE-NC Stage 3. Key findings include the identification of features of TDP-43 histomorphology and severity in the MFG, supported by clinical information, that reliably differentiated severe LATE-NC from FTLD-TDP. These findings provided the basis for a diagnostic rubric for LATE-NC Stage 3, as depicted schematically in Fig. 9. We also surveyed the NACC/ADGC cohort in order to shed light on the genetic factors influencing susceptibility to LATE-NC Stage 3. These findings implicated the GRN rs5848 homozygosity as a genetic contributor to LATE-NC Stage 3.

Fig. 9.

Diagnostic rubric for differentiating LATE-NC Stage 3 from other TDP-43-opathies. Criteria denoted with red boxes are potential pitfalls highlighted in the current study. The workflow begins with an assessment of TDP-43 immunohistochemistry in the middle frontal gyrus (MFG). If more than 15 TDP-43+ lesions per high-power field (HPF) are detected in the MFG, specific pathological and genetic criteria are evaluated to classify the case as FTLD-TDP types B, C, D, or E. Additional genetic markers (GRN or C9ORF72 variants) and a history of motor neuron disease are used for further classification. Rare cases with Alzheimer’s disease (AD)-type clinical history, TMEM106B risk alleles, and severe ARTAG are a separate category. Residual cases are assigned to LATE-NC Stage 3

LATE-NC Stage 3 pathology affects ~ 4% of individuals over 85 years of age [49]. This prevalence is markedly higher than that of FTLD-TDP, which has a lifetime risk of approximately 1:1000 [10, 35]. LATE-NC Stage 3 represents the most advanced stage of TDP-43 proteinopathy in existing staging frameworks, including the 0–6 scale proposed by Josephs et al.[28] and the 0–5 scale introduced by Nag et al. [45]. In a prior study comparing Josephs TDP-43 stages 1–3 (“limbic”) vs stages 4–6 (“diffuse”), Carlos et al. [6] reported that the diffuse TDP-43 stages showed more severe clinical/cognitive impairment, a higher frequency of TMEM106B and APOE risk alleles, and less-widespread amyloid pathology [6]. For reasons that were articulated previously [51, 52], the adoption of a simplified 0–3 scale for LATE-NC staging in routine autopsy diagnoses was decided upon based on considerations of practicality, generalized utility, and diagnostic clarity.

Prior studies provided partial guidance as to the delineation of diagnostic "boundary zones" between severe LATE-NC and early FTLD-TDP. In the first place, FTLD-TDP and other non-LATE-NC pathologic entities are relatively uncommon in elderly populations studied thus far. For example, in a study encompassing 13 community-based cohorts and 6,251 brains, none of these cases displayed fully developed FTD/FTLD-TDP [49]. When clear-cut FTD/FTLD is diagnosed, that diagnosis takes precedent: brains with specific subtypes of FTLD-TDP, such as those with the qualitative pathological patterns of FTLD-TDP Types C and D, should not be diagnosed as LATE-NC [52]. Robinson et al. [65] addressed the issue of differentiating severe LATE-NC from FTLD-TDP (Types A and/or B) by generating a “hand-counting” method for assessing TDP-43 pathology in the MFG. This method successfully distinguished LATE-NC from FTLD-TDP types A and B with more than 98% accuracy across two large autopsy cohorts: one community-based (with few FTLD-TDP cases) and one clinic-based (enriched for FTLD-TDP cases). The reliability of this method was confirmed by blinded evaluations of five independent neuropathologic raters [65], and was also useful in the present study (Fig. 4). The finding that the severity of MFG TDP-43 proteinopathy can be used to distinguish between LATE-NC and FTLD-TDP was further corroborated by Young et al. [86]. These investigators employed a sophisticated “SuStain” methodology integrating semi-quantitative ratings of TDP-43 across a large number of brain regions with machine learning of spatial progression patterns. While this approach confirmed distinct patterns of TDP-43 pathology among TDP-43-opathies, it also raised a question as to whether there was some overlap between late-stage LATE-NC and early-stage FTLD-TDP, and the possibility of shared etiologies between these conditions [86]. Additional evidence supporting some overlap between FTLD-TDP and LATE-NC Stage 3 was presented by Josephs et al. [5], who analyzed brains from the Mayo Clinic brain bank—one of the largest (if not the largest) collections of FTD/FTLD brains in the world. This study demonstrated that when using digital pathologic assessments of TDP-43 proteinopathy, the results for some FTLD-TDP cases, including genetically driven ones, partially overlapped with those of LATE-NC [5]. Seen in light of the current work, these findings highlight the challenges involved with distinguishing TDP-43-opathies based on pathology alone (particularly in a single brain region), and underscore the need for continued refinement of diagnostic criteria and methodologies.

The present study elucidated several factors that may have led to diagnostic challenges in prior studies, while resolving the diagnostic ambiguities. We provided a data-driven set of criteria to effectively differentiate LATE-NC from FTLD-TDP based on clinical and neuropathological observations (Fig. 9). One diagnostically challenging scenario arises from FTLD-TDP cases with relatively low levels of TDP-43 pathology in the MFG, where evaluation of the MFG alone constitutes a potentially misleading context for diagnosis. The low amount of detected TDP-43 pathology levels in some FTLD-TDP type B cases was related to the presence of gNCIs, which were in several cases the predominating TDP-43 inclusion type. These gNCI lesions were prone to undercounting by digital methods, but a trained neuropathologic can readily identify these lesions. Consequently, gNCI lesions should be recognized as a morphologic point of differentiation between LATE-NC and FTLD-TDP type B. Furthermore, cases of motor neuron disease (FTLD-MND or ALS) may exhibit relatively low TDP-43 pathology levels in the MFG. A clinical history compatible with motor neuron disease should prompt neuropathologic evaluation of motor neuron regions (primary motor cortex and spinal cord) and TDP-43 pathologic positivity in motor neurons would steer a neuropathologist toward a primary diagnosis of FTLD-MND or ALS.

Two cases in the present study stood out because they had a clinical history compatible with LATE-NC, but had severe TDP-43 proteinopathy in the MFG. One was a person from the UK-BB cohort who died at age 91 with a diagnosis of “Probable AD”, and the second individual was from the Mayo BB with a clinical diagnosis of AD/PD, dying at age 79 years of age. Each had high density of MFG TDP-43 pathology. Upon careful analyses, both of these cases had unusually high levels of tau astrogliopathy in their hippocampi. The UK-BB case was confirmed to be TMEM106B risk allele homozygous. We consider that both of these cases likely represent limbic-predominant neuronal inclusion body 4R tauopathy (LNT)—a recently-recognized non-FTLD, non-LATE-NC condition with a clinical presentation similar to LATE-NC, yet the same brains have a set of pathologic and genetic features that set them apart [1, 37–39, 68]. A similar (if not identical) entity has also been referred to as “temporal-predominant neuro-astroglial tauopathy” [39]. Taken together, these findings emphasize the importance of integrating neuropathological criteria, along with thorough review of clinical history (and when possible, genetics), to achieve accurate differential diagnoses, as would be the case in most routine autopsy practices. We also underscore the necessity, in some cases, of applying “descriptive” diagnoses, particularly in cases with unusual or ambiguous pathologies [47].

Based on the clinical-pathological correlation results in prior published studies [6, 50, 83], we expected to find substantial differences in cognitive performance when comparing between LATE-NC Stages 2 and 3; however, we did not find a particular neurocognitive test or neuropsychiatric symptom that could differentiate LATE-NC Stages 2 or 3, even at the group level. This may be due to the NACC NP Data Set cohort being studied [3], in which almost 90% of analyzed individuals had dementia prior to death—a far higher dementia prevalence than would be expected in a community-based sample [21]. The majority of LATE-NC cases were diagnosed clinically as “Probable Alzheimer’s Disease,” highlighting the limitations in the specificity of clinical diagnoses relative to underlying neuropathology [24, 41]. These results also highlight the need for refined diagnostic frameworks that incorporate both clinical (including, in the future, biomarkers) and pathological criteria to improve the accuracy of distinguishing LATE-NC from other neurodegenerative conditions.

In addition to findings related to pathology and clinical symptomatology, the current study found that a GRN genetic variant (the SNV rs5848 risk allele) was associated with increased risk for LATE-NC Stage 3. This finding was unexpected and needs to be confirmed in other autopsy series. Nonetheless, there are several possible implications that are worthy of discussion. In autopsy cohorts studied to date, the presence of the rs5848 risk allele conferred a ~ 1.3-fold increased odds ratio for LATE-NC [13, 43], an observation that has been replicated by several groups [14, 30–32, 54]. The rs5848 risk allele is linked with lower blood progranulin levels [61, 62, 75] and with increased inflammatory mediators in cerebrospinal fluid [16]. The multifaceted involvement of GRN in animal biology is remarkable [60]. GRN and its cognate protein progranulin (characterized as an adipokine) play many different roles in humans including in neuroinflammation, cell growth, and wound repair [15]. GRN mutations are correspondingly pleiotropic in terms of their impact on human disease. For example, the same rs5848 GRN gene variant has been associated with altered risk for LATE-NC, FTLD-TDP, ADNC/tauopathy, Parkinson disease, multiple sclerosis, and Gaucher disease [8, 13, 14, 20, 26, 50, 63, 72, 73]. GRN/progranulin also plays key roles in cancer biology [78, 87]. While this pleiotropy underscores that there are aspects of pathogenetic overlap between those diseases, perhaps related to inflammation, it does not imply that those diseases are all essentially the same. Another reason for interest in the rs5848 genetic variant is that the risk allele is far more commonly inherited by persons of African ancestry (~ 75%) than by persons of European ancestry (~ 30%) [30]. There is exciting potential for clinical therapies to counteract decreased progranulin levels [17, 64, 66]. Since persons of African ancestry may be relatively likely to require that therapy, future optimized LATE-NC prevention strategies could differ based on the populations served.

The present study had several limitations that warrant consideration. We note that although only 51 different cases were evaluated in detail, we included each relevant LATE-NC Stage 3 case from both the UK-BB and UCI-BB cohorts in the present study, comprising over 950 community-dwelling individuals as a denominator. However, the ethnoracial diversity of the cohorts was limited, highlighting a critical area for future investigation. Although we found that all LATE-NC Stage 3 cases could be classified confidently with the proposed diagnostic rubric (Figs. 4 and 9), there is a possibility that a cohort of people representing true diagnostic ambiguity between LATE-NC and FTLD-TDP may exist but was not captured in this study. A further limitation of the present article relates to the study design, wherein tissue sections were stained immunohistochemically for phosphorylated TDP-43 at the UK-ADRC after being received from external institutions. Therefore, variability in tissue fixation and storage practices may have influenced the staining characteristics. Previous research has demonstrated that TDP-43 immunohistochemistry is susceptible to technical variability due to these factors [22, 85]. Future studies need to optimize immunohistochemical techniques to ensure consistent and reliable results when integrating tissue samples from multiple institutions. These refinements would enhance the robustness of subsequent investigations.

In conclusion, our findings indicate that LATE-NC Stage 3 cases from three large autopsy cohorts could be reliably distinguished from FTLD-TDP. The classification rubric we generated was based on a set of diagnostic criteria, including qualitative neuropathologic assessments (FTLD-TDP and NCI subtypes), measures of TDP-43 pathology in the MFG, and integrating the pathologic information with clinical data. For the purposes of routine diagnostic practice, the “hand-counting” method of Robinson et al. [86] appears to be robustly useful, whereas more quantitative digital pathology is not required. In addition, the identification of an association between the GRN risk allele (SNV rs5848) and LATE-NC Stage 3 risk suggests that genetic factors may influence the progression of LATE-NC from Stage 2 to Stage 3.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

ADGC

Alzheimer’s Disease Genetics Consortium

ADNC

Alzheimer’s disease neuropathologic changes

ADRC

Alzheimer’s disease research center

ALS

Amyotrophic lateral sclerosis

ARTAG

Age-related tau astrogliopathy

CBD

Corticobasal degeneration

CDR

Clinical Dementia Rating Scale

FTD

Frontotemporal degeneration

FTLD-MND

Frontotemporal lobar degeneration with motoneuron disease

FTLD-TDP

Frontotemporal lobar degeneration with TDP-43 inclusions

LATE-NC

Limbic-predominant age-related TDP-43 encephalopathy neuropathologic changes

LNT

Limbic-predominant neuronal inclusion body 4R tauopathy (or, alternatively, limbic neuro-astroglial tauopathy)

Mayo-BB

Mayo Clinic ADRC brain bank

MFG

Middle frontal gyrus

MINT

Multilingual Naming Test

MMSE

Mini Mental State Examination

MoCA

Montreal Cognitive Assessment

NACC

National Alzheimer’s Coordinating Center

NCI

Neuronal cytoplasmic TDP-43 inclusion (“gNCI” indicates “granular NCI”, “dNCI” indicates “diffuse NCI”)

NPI-Q

Neuropsychiatric Inventory—Quantitative

NPS

Neuropsychiatric symptoms

SNP/SNV

Single nucleotide polymorphism/variant

UCI-BB

The 90+ Study and University of California Irvine ADRC brain bank

UDS

Uniform Data Set

UK-BB

University of Kentucky ADRC brain bank

WMS-R

Wechsler Memory Scale-Revised Logical Memory

Author contributions

RKS and PTN co-wrote the main manuscript text and conceived of the project. Statistics, analyses and figures were prepared by YK, KZA, XW, & DWF. Experiments were performed by PP, AMN, JRCA, & RKS. Tissue was provided with consultations by DWD, NBG, MMC, CHK, SAS, DCW, SAB, and TJM. Clinical consultation was provided by GAJ. All authors reviewed the manuscript.

Funding

NIH grants R01 AG061111, R01 AG057187, P30 AG072946, RF1 NS118584, T32 AG078110, R01 AG021055, P30 AG066519, P30 AG062677, RF1 AG082339.

Declarations

Conflict of interest

The authors declare no competing interests.