FTLD‐TDP versus LATE‐NC: Experience of a Brain Bank specializing in FTLD‐TDP

Abstract

INTRODUCTION

Similarities between frontotemporal lobar degeneration with transactive response DNA‐binding protein of 43 kDa (TDP‐43) (FTLD‐TDP) and limbic‐predominant age‐related TDP‐43 encephalopathy neuropathologic change (LATE‐NC) raise questions about whether they represent distinct entities or a single disease spectrum. The literature mostly examined series with disproportionate numbers of LATE‐NC over FTLD‐TDP.

METHODS

Leveraging a clinicopathological collection of FTLD‐TDP (N = 148) from the University of California, San Francisco, we compared demographic, clinical, genetic, and neuropathological features of FTLD‐TDP, particularly FTLD‐TDP type A (N = 39), and LATE‐NC (N = 42).

RESULTS

FTLD‐TDP type A cases were younger at onset and death, had shorter disease duration, and frequent genetic causes (GRN, C9ORF72) compared to LATE‐NC, which were mostly sporadic and older. Blinded evaluation of middle frontal gyrus (MFG) TDP‐43 immunostaining alone proved insufficient to reliably differentiate FTLD‐TDP type A from LATE‐NC stage 3. However, factoring in all neuropathologic features, FTLD type A and LATE‐NC could be differentiated with >95% confidence.

DISCUSSION

These overall findings support distinct diagnostic entities for FTLD‐TDP and LATE‐NC.

Highlights

• FTLD‐TDP and LATE‐NC can overlap clinically and pathologically, but key features help distinguish them in practice.

• Using brain bank data enriched for FTLD‐TDP, we provide an original side‐by‐side comparison of FTLD‐TDP and LATE‐NC cases.

• Age at death and regional vulnerability patterns differ between FTLD‐TDP and LATE‐NC in our cohort.

• Assessing TDP‐43 pathology in the MFG alone is not sufficient to differentiate FTLD‐TDP from LATE‐NC.

• We offer a perspective on the literature comparing FTLD‐TDP and LATE‐NC, highlighting points of convergence/divergence and implications for classification and interpretation across studies.

1. INTRODUCTION

Abnormally misfolded and aggregated transactive response DNA‐binding protein of 43 kDa (TDP‐43) is a major pathological protein in frontotemporal lobar degeneration (FTLD) and motor neuron disease (MND). Most individuals showing a FTLD histopathological pattern (fronto‐insulo‐temporal‐predominant atrophy with vacuolation, astrogliosis, and neuronal loss) with TDP‐43 inclusions (FTLD‐TDP) present with behavioral variant frontotemporal dementia (bvFTD) or a primary progressive aphasia (PPA), with or without MND. FTLD‐TDP is subcategorized into at least four subtypes (harmonized as types A to D) based on the pattern of morphological features, subcellular localization, and laminar distribution of TDP‐43 inclusions ¹ .

Once specialized brain banks routinely screened for TDP‐43 proteinopathy, several groups reported abnormal TDP‐43 deposits in the brains of older individuals, yet these individuals lacked FTLD histopathological patterns or FTD clinical syndromes. Instead, an amnestic syndrome was often observed, even without significant accompanying Alzheimer's disease (AD). ² A workgroup ultimately named this entity “limbic‐predominant, age‐related TDP‐43 encephalopathy neuropathologic change” (LATE‐NC). ³ In contrast to FTLD‐TDP, in which cerebral cortical pathology predominates, LATE‐NC primarily affects limbic regions, reaching the neocortex in only 11% to 13% of the cases. ² , ⁴ While the lifetime risk of developing FTLD‐TDP is approximately 1:1000, ⁵ the prevalence of LATE‐NC increases linearly with age, reaching 1:2 in some 90+ series. ² , ³ FTLD‐TDP is rare overall and occurs in 31% of late‐onset (i.e., symptoms starting after 65 years of age) cases across the FTD clinical spectrum and in 44% of early‐onset FTD cases. ⁶ Lastly, whereas many FTLD‐TDP cases have a monogenic origin, such as a C9ORF72 expansion or GRN or TARDBP mutations, LATE‐NC is sporadic.

Though FTLD‐TDP and LATE‐NC exhibit distinct characteristics, they also share some features, especially regarding TDP‐43 pathomorphology. This tension has led to an ongoing debate over whether FTLD‐TDP and LATE‐NC represent a continuous spectrum of the same underlying pathology or discrete pathological entities with notable commonalities. Advocates of the former perspective emphasize the distinct clinical and neuroimaging manifestations in FTLD‐TDP and LATE‐NC. ² , ³ , ⁷ A recent study analyzed semi‐quantitative scores for TDP‐43 proteinopathy across 21 brain regions and basic clinical and genetic data using the subtype and stage inference algorithm to classify previously defined FTLD‐TDP and LATE‐NC, and their results corroborated the initial diagnoses assigned at autopsy. ⁸ Proponents of the “continuum” viewpoint underscore shared vulnerability within limbic regions and some shared genetic risk. ⁹ , ¹⁰ These authors draw parallels between FTLD‐TDP and LATE‐NC and early‐ and late‐onset AD, respectively, wherein early‐onset AD cases are more infrequent and display a disproportionately higher burden of neocortical pathology and behavioral symptomatology. ¹¹ In addition, LATE‐NC inclusions in some patients resemble those seen in FTLD‐TDP type A. ¹² FTLD‐TDP type A features limbic pathology, and 8% to 20% of LATE‐NC show neocortical TDP‐43 deposits. ² , ¹³ Although the lack of ante mortem biomarkers for TDP‐43 pathology is an obstacle for discerning the pattern of anatomical spreading in FTLD‐TDP type A over time, based on cross‐sectional neuroimaging and pathological data, FTLD‐TDP type A's involvement in peri‐allocortical (“paralimbic”) transition zones prior to spreading into neocortical and limbic structures has been widely replicated. ¹⁴ , ¹⁵

Only a handful of clinicopathological studies have sought to determine whether FTLD‐TDP and LATE‐NC are part of the same spectrum or different entities. These studies used either databases such as the National Alzheimer's Coordinating Center (NACC) clinicopathological cohort, which provides standardized data collection but limited granularity, or smaller cohorts from individual centers that offer more comprehensive data but are relatively underpowered, especially for FTLD‐TDP type A. With these caveats in mind, it was found that FTLD‐TDP had greater hippocampal and neocortical involvement, greater clinical severity at a younger age, and less of a relationship with AD neuropathologic change (ADNC) and the apolipoprotein E gene (APOE), than LATE‐NC. ¹¹ , ¹⁶ , ¹⁷ , ¹⁸

Shahidehpour et al. ¹⁹ proposed a pragmatic rubric that highlights middle frontal gyrus (MFG) pathology to help distinguish severe LATE‐NC (stage 3) from FTLD‐TDP. This emphasis fits the 2019 workgroup for LATE‐NC sampling recommendations – amygdala, hippocampus, and MFG – recognizing that the amygdala and hippocampus are commonly affected in both LATE‐NC and FTLD‐TDP type A and, thus, less discriminative. The rubric sensibly incorporated available clinical/genetic information (e.g., MND, mutations) when assigning LATE‐NC stage 3. A constructive next step is to test how well MFG‐weighted histopathology alone separates LATE‐NC from FTLD‐TDP type A across larger, more diverse cohorts, particularly with increased representation of type A cases. However, given the limited number of FTLD‐TDP type A cases available to Shahidepour et al., it remains unclear whether MFG pathology alone reliably differentiates these entities across all FTLD‐TDP subtypes versus LATE‐NC.

The University of California, San Francisco (UCSF) Neurodegenerative Disease Brain Bank (NDBB) specializes in FTD and has one of the largest available FTLD‐TDP clinicopathological collections, with nearly 150 cases. Here, we critically appraise the literature, focusing on whether FTLD‐TDP and LATE‐NC represent similar or distinctive neuropathological processes, and discuss these findings through the lens of our experience with the UCSF NDBB cohort. Moreover, we tested whether, in cases positive for TDP‐43 pathology in the MFG, the evaluation of this single neocortical region is sufficient to differentiate FTLD‐TDP type A from LATE‐NC stage 3.

RESEARCH IN CONTEXT

1. Systematic review: We conducted a comprehensive literature review spanning 2010 through 2024, focusing on clinicopathological studies exploring TDP‐43 proteinopathies. Specifically, we identified studies that compared the neuropathological characteristics of FTLD‐TDP and LATE‐NC, many of which offered insights into clinical profiles, demographic information, and genetic factors.

2. Interpretation: The UCSF NDBB data (148 FTLD‐TDP cases) suggest that LATE‐NC is distinct from FTLD‐TDP, including type A, although the two share some key clinical features.

3. Future directions: Current efforts to characterize the FTLD‐TDP subtypes and LATE‐NC should expand beyond clinical characteristics and include genetic, gene expression, and other molecular features to contrast these two entities. Further, elucidating the protein structure of TDP‐43 aggregates in LATE‐NC by cryo‐electron microscopy would allow for direct comparison to the FTLD‐TDP type A‐associated chevron TDP‐43 fold and other FTLD‐TDP folds. Also, evaluating a single neocortical region, the MFG, is insufficient for reliable differentiation because some FTLD‐TDP type A cases may have less disease burden in that structure than anticipated. Of note, insufficient single‐region discrimination does not contradict broader clinicopathological differences supporting distinct disease entities.

2. METHODS

We planned to summarize first all FTLD‐TDP and LATE‐NC cases in the UCSF NDBB, then to focus our comparison between type A and LATE‐NC cases. The UCSF NDBB provides autopsy services to all patients and research participants at the Edward and Pearl Fein Memory and Aging Center. The clinical population is a tertiary referral center including a mix of San Francisco Bay Area community members and referrals from across the state and country. Research participants enter through many programs, including the Alzheimer's Disease Research Center (ADRC), that span healthy aging cohorts and participants with FTD or atypical AD syndromes. We searched the UCSF NDBB to identify all cases from 2005 to 2022 that had TDP‐43 inclusions. The NDBB has used the same antibody (10782‐2‐AP, ProteinTech, RRID:AB_615042) over time with minimal protocol changes along with careful controls to ensure staining development is consistently applied. Cases were designated TDP‐43‐positive if TDP‐43‐immunoreactive neuronal cytoplasmic inclusions, dystrophic neurites, neuronal intranuclear inclusions, and/or gray matter threads were identified in any area examined. Cases with isolated perivascular deposits were omitted. At a minimum, all cases were examined for TDP‐43 in the amygdala, hippocampal formation, anterior cingulate gyrus, inferior temporal gyrus, and MFG. Cases with a primary neuropathological diagnosis of amyotrophic lateral sclerosis (ALS)‐TDP were excluded. Further, seven cases were excluded from analysis because they had a primary neuropathological diagnosis other than FTLD‐TDP, specifically: Four had corticobasal degeneration, one Lewy body disease, one vascular brain injury, and one chronic traumatic encephalopathy with a clear contact sports history. These 190 cases were first categorized into FTLD‐TDP or LATE‐NC based on the prospective neuropathological assessment, which meets currently published standards; as diagnostic criteria evolved, cases were re‐reviewed over time to maintain consistency with updated nomenclature and classification. FTLD‐TDP cases were subtyped according to the harmonized scheme from Mackenzie et al. ¹ Briefly, FTLD‐TDP type A exhibits many neuronal cytoplasmic inclusions, which are often compact and crescentic, and short dystrophic neurites, predominantly of cortical layer 2; type B shows moderate cytoplasmic inclusions, which may be speckled or skein‐like, and with fewer neurites affecting all cortical layers; type C has many, long, ropy dystrophic neurites predominantly in layer 2 with few cytoplasmic inclusions; and type D has many short dystrophic neurites and lentiform neuronal intranuclear inclusions affecting all cortical layers. To ensure consistency and greater inter‐rater reliability in applying these classifications, all cases’ diagnoses and subtypes are reviewed in a neuropathology consensus conference. Of note, FTLD‐TDP type A and LATE‐NC cases often have compact neuronal cytoplasmic inclusions across most hippocampal subfields and the dentate gyrus, and they both can have many small threads, particularly in the subiculum and CA1. Only when exiting the hippocampus to other neocortical areas and also the ventral striatum do these diagnoses diverge in their relative involvement of structures; 21 cases were unclassifiable (U), many due to a mixture of type A and type B features with others too sparse to classify or a mix of features atypical for both types A and B. Cases designated prospectively or retrospectively as LATE‐NC were classified as limbic (stages 1 or 2) or neocortical (stage 3, defined by the presence of TDP‐43 inclusions in the MFG). Current neuropathological criteria for LATE‐NC do not formally incorporate inclusion morphology as part of the diagnostic framework.

First, cases were compared by demographics, brain weight, and proportion with neocortical TDP‐43 deposits using the diagnoses in the neuropathological reports (Table 1). FTLD‐TDP cases with intermediate to high ADNC ²⁰ were classified as “FTLD‐TDP+AD” regardless of subtype. Thus, initial comparisons included the following five groups: LATE‐NC, FTLD‐TDP types A, B, C, and U, and FTLD‐TDP (any type)+AD. All LATE‐NC cases also had AD present, so the LATE‐NC group was not divided into subgroups of with and without AD and is referred to as “LATE‐NC+AD” hereafter.

TABLE 1.

Demographics.

	LATE‐NC	FTLD‐TDP +AD	FTLD‐TDP type A	FTLD‐TDP type U	FTLD‐TDP type B	FTLD‐TDP type C
N	42	18	36	18	43	33
Mean age at onset ± SD, in years (range)	68.2 ± 10.4 (49 to 86)	64.0 ± 8.6 (44 to 75)	60.9 ± 6.9 (47 to 81)	57.3 ± 8.5 (42 to 70)	53.3 ± 10.4 (18 to 71)	59.1 ± 9.2 (44 to 88)
Disease duration, mean ± SD, in years (range)	11.1 ± 4.5 (4 to 24)	10.3 ± 4.5 (3 to 23)	7.7 ± 3.2 (3 to 21)	7.8 ± 5.5 (1 to 30)	8.0 ± 6.0 (1 to 30)	13.2 ± 4.1 (4 to 21)
Mean age at death ± SD, in years (range)	79.6 ± 9.2 (58 to 98)	73.9 ± 5.8 (65 to 85)	68.6 ± 7.7 (52 to 84)	65.1 ± 7.9 (41 to 98)	61.3 ± 9.4 (41 to 78)	72.5 ± 6.7 (62 to 92)
Female sex, n (%)	21 (50%)	7 (39%)	23 (64%)	6 (33%)	19 (44%)	16 (48%)
Mean brain weight (g) ± SD (range)	1101 ± 144 (836 to 1441)	1100 ± 166 (847 to 1423)	1027 ± 193 (443 to 1540)	1085 ± 206 (443 to 1540)	1210 ± 146 (962 to 1510)	1019 ± 140 (790 to 1262)
Neocortical TDP‐43 proteinopathy	2.4%	100%	100%	100%	100%	100%

Next, to interrogate whether FTLD‐TDP and LATE‐NC represent a spectrum of disease, we excluded cases meeting the criteria for FTLD‐TDP types B or C because FTLD‐TDP type A and LATE‐NC show the strongest morphological overlap of TDP‐43 inclusions. We then narrowed our analyses to focus on the FTLD‐TDP type A, U, A/U+AD, and LATE‐NC+AD groups.

In parallel, three blinded raters (DLF, SS, LTG) assessed the MFG in all type A and LATE‐NC Stage 3+AD cases to evaluate whether routine diagnostic examination of this region alone – commonly the only neocortical region assessed for TDP‐43 in many brain banks and emphasized in a published diagnostic rubric ¹⁹ – is sufficient to differentiate type A from LATE‐NC Stage 3. Raters assessed digitized slides with images taken at 20× on a Zeiss Axioscan 7 microscope. Raters were asked the following question: “Is this case FTLD‐TDP type A or LATE‐NC stage 3?” The rates of agreement among raters and with an original neuropathological diagnosis were analyzed. Raters provided semi‐quantitative scores, ranging from 0 to 3, based on the hematoxylin and eosin (H&E) and TDP‐43‐stained slides for vacuolation, gliosis, neuronal loss, neuronal intranuclear inclusions, neuronal cytoplasmic inclusions, dystrophic neurites, glial cytoplasmic inclusions, gray matter threads, and white matter threads. Raters knew that most cases were diagnosed as FTLD‐TDP type A, but they were unaware of the exact sample size of each diagnosis.

Most analyses were descriptive. We used ANOVA to compare continuous variables with Sidak tests for post hoc comparisons, except for one comparison for which a t‐test was used: FTLD‐TDP+AD versus all FTLD‐TDP with low ADNC groups combined. A Pearson's chi‐squared test was used for categorical variables. Agreement analyses were conducted using Fleiss’ kappa with customary interpretation of intermediate values. ²¹ , ²² All analyses were conducted on IBM SPSS statistics version 27 or STATA version 18. Results were statistically significant for p < 0.05.

To contextualize our findings, we searched the literature from 2010 to 2025, focusing on clinicopathological studies that explored TDP‐43 proteinopathies. We identified studies that compared the neuropathological characteristics of FTLD‐TDP and LATE‐NC, many of which offered insights into clinical profiles, demographic information, and genetic factors. We organized the competing arguments pertaining to the question of whether FTLD‐TDP and LATE‐NC should be on a neuropathological spectrum or considered distinct entities.

3. RESULTS

3.1. UCSF NDBB: All FTLD‐TDP and LATE‐NC cases

Cases with TDP‐43 proteinopathy (N = 190) in the initial analysis were ascertained from the UCSF NDBB from 2005 to 2022 (female 48.2% (N = 92); mean age at death of 70.3 ± 10.5 years with range of 41 to 98 years). Table 1 depicts the demographics and brain weight by disease group (namely, LATE‐NC+AD, FTLD‐TDP+AD, and FTLD‐TDP types A, B, C, and U). Except in comparison to FTLD‐TDP+AD, the LATE‐NC+AD group had older ages at symptom onset (F(5, 181) = 12.0, vs. type A, p = 0.011; type B, p < 0.001; type C, p = 0.001; and type U, p = 0.001) and death (F(5, 184) = 24.2, vs. type A, p < 0.001; type B, p < 0.001; type C, p = 0.004; and type U, p < 0.001). Compared to the combined FTLD‐TDP cases with none or low ADNC, the FTLD‐TDP+AD group was older at death (t(146) = 3.25, p < 0.001). FTLD‐TDP+AD was older at age of onset than type B (p = 0.001) and at age of death than types B (p < 0.001) and U (p = 0.021). Disease duration was shorter in type A versus LATE‐NC+AD (F(5, 181) = 7.40, p = 0.024), types A versus C (p < 0.001), LATE‐NC+AD versus type B (p = 0.034), types B versus C (p < 0.001), and types C versus U (p = 0.002). Brain weight was lower in type A versus B (F(5, 180) = 6.95, p < 0.001), LATE‐NC+AD versus type B (p = 0.035), and type C versus B (p < 0.001). There was no difference in sex distribution across groups.

3.2. UCSF NDBB: FTLD‐TDP types A, U, or A/U+AD versus LATE‐NC+AD

Whereas all type A and U cases had frontal (at least MFG and inferior frontal gyrus) neocortical TDP‐43 proteinopathy, only one (2.4%) LATE‐NC+AD case showed pathology in the MFG (i.e., LATE‐NC stage 3 by definition). The remaining LATE‐NC+AD cases had TDP‐43 pathology restricted to limbic and paralimbic areas. All LATE‐NC cases had at least intermediate ADNC levels (hence called “LATE‐NC+AD” throughout), whereas only five (12.2%) type A and two (2.0%) type U cases had ADNC intermediate or greater (Table 2).

TABLE 2.

Diagnostic data.

	LATE‐NC	FTLD‐TDP types A/U +AD	FTLD‐TDP type A	FTLD‐TDP type U
Genetics
C9ORF72, n (%) a	0/35 (0%)	3/7 (42.9%)	5/32 (15.6%)	13/14 (92.9%)
GRN, n (%) a	0/35 (0%)	2/7 (28.6%)	22/32 (68.8%)	0/14 (0%)
TARDBP, n (%) a	0/35 (0%)	0/6 (0%)	0/30 (0%)	1/14 (7.1%)
Percentage sporadic, n (%) a	35/35 (100%)	2/7 (28.6%)	5/32 (15.6%)	0/14 (0%)
Percentage unknown genetic status	7/42 (16.7%)	0/7 (0%)	4/36 (11.1%)	4/18 (22.2%)
APOE ε4 carrier	22/35 (62.9%)	5/6 (83.3%)	4/30 (13.3%)	6/15 (40.0%)
TMEM106B rs1990622 A/A	17/35 (48.6%)	3/6 (50.0%)	19/30 (63.3%)	7/15 (46.7%)
Clinical Diagnoses, n (%)
AD‐type dementia or PCA b	32/42 (76.2%)	3/7 (42.9%)	1/36 (2.8%)	0/18 (0%)
bvFTD	4/42 (9.5%)	1/7 (14.3%)	22/36 (61.1%)	7/18 (38.9%)
bvFTD‐MND or ALS	0/42 (0%)	0/7 (0%)	0/36 (0%)	9/18 (50.0%)
CBS/PSPS	2/42 (4.8%)	0/7 (0%)	4/36 (11.1%)	1/18 (5.6%)
nfPPA	0/42 (0%)	1/7 (14.3%)	4/36 (11.1%)	0/18 (0%)
svPPA	0/42 (0%)	0/7 (0%)	1/36 (2.8%)	1/18 (5.6%)
AD versus FTD	2/42 (4.8%)	0/7 (0%)	1/36 (2.8%)	0/18 (0%)
Other	2/42 (4.8%)	2/7 (28.6%)	3/36 (8.3%)	0/18 (0%)
ADNC status (N, %)
Not	0/42 (0%)	0/7 (0%)	15/36 (41.7%)	3/18 (16.7%)
Low	0/42 (0%)	0/7 (0%)	21/36 (58.3%)	15/18 (83.3%)
Intermediate	3/42 (7.1%)	4/7 (57.1%)	0/36 (0%)	0/18 (0%)
High	39/42 (92.9%)	3/7 (42.9%)	0/36 (0%)	0/18 (0%)

^aPercentage of cases of known genetic status.

^bNo logopenic variant PPA cases for any group.

In the LATE‐NC+AD group, 83% of cases underwent genetic testing, and none harbored a FTLD‐related mutation or expansion. Further, no LATE‐NC+AD cases had a family history of early‐onset dementia or a first‐degree relative with a FTLD‐related clinical syndrome. In contrast, a genetic origin was more common in the FTLD‐TDP groups. In the type A (without AD) group, 32 of 36 cases had available genotyping, and GRN mutations (68.8%) and C9ORF72 expansions (15.6%) were common. In the type U (without AD) group, 14 out of 18 cases had available genotyping; 92.9% had a C9ORF72 expansion, and one case had a TARDBP mutation. In the FTLD‐TDP types A/U+AD group, 17 of 18 cases had available genotyping with 29% having a C9ORF72 expansion and 11.8% a GRN mutation. Of note, APOE ε4 carriers were more prevalent in the LATE‐NC+AD (62.9%) and FTLD‐TDP type A/U+AD (83.3%) groups compared to the type A (13.3%) and type U groups (40%; χ²[3, N = 86] = 20.4, overall p < 0.001).

Most LATE‐NC+AD cases had a clinical diagnosis of AD‐type dementia (76.2%, Table 2), yet four (9.5%) had a clinical diagnosis of bvFTD, and another two had a diagnosis of AD versus FTD (4.8%). The proportion of cases with AD‐type dementia or AD versus FTD clinical diagnoses was much lower in type A (5.6%) and type U (0%) groups (Table 2) compared with LATE‐NC+AD.

3.3. LATE‐NC Stage 3 versus FTLD‐TDP type A in cases with moderate to high ADNC

As expected from the literature, ²³ , ²⁴ , ²⁵ , ²⁶ , ²⁷ , ²⁸ all UCSF NDBB cases of LATE‐NC coexisted with ADNC (“LATE‐NC+AD”). Therefore, LATE‐NC+AD and FTLD‐TDP type A comparisons should minimize the confound of co‐present ADNC. Hence, we compared cases with at least intermediate ADNC that displayed inclusions with type A‐like morphology in the MFG. We identified a total of eight cases meeting these specific criteria: one LATE‐NC stage 3 (+AD), five FTLD‐TDP type A, and two FTLD‐TDP type U (Table 3). Of note, two patients in this group had GRN mutations (both with FTLD‐TDP type A), and three had C9ORF72 expansions (one type A and two type U). The remaining three sporadic cases exhibited similar ages at onset, ranging from 72 to 75 years, and they reached an age at death in the ninth decade, spanning from 81 to 86 years (Table 3).

TABLE 3.

LATE‐NC Stage 3 or FTLD‐TDP type A or U with intermediate or high ADNC.

Case no.	TDP‐related diagnosis	Sex	Age of onset (years)	Disease duration (years)	Genetics	Clinical diagnosis
1	LATE‐NC Stage 3	F	72	14	Sporadic	bvFTD
2	Type A	F	75	7	Sporadic	AD
3	Type A	M	75	8	Sporadic	PPA, unspecified
4	Type A	M	62	10	GRN	AD
5	Type A	M	69	5	GRN	nfvPPA
6	Type A	M	54	13	C9ORF72	bvFTD
41	Type U	F	65	13	C9ORF72	unspecified
42	Type U	M	56	14	C9ORF72	AD

The one patient with LATE‐NC stage 3 (+AD) had a clinical diagnosis of bvFTD. She presented with short‐term memory loss and quickly developed behavioral symptoms, including stereotypical behaviors, paranoid ideation, increased libido, changes in social conduct, and progressive difficulties with task execution, warranting a clinical diagnosis of bvFTD. Family history included late‐onset dementia in two siblings; testing for FTLD‐related genes was negative. TDP‐43 proteinopathy in neocortical areas was moderate. In addition to comorbid high ADNC, the brain exhibited severe atherosclerosis, amygdala‐predominant Lewy body disease, no hippocampal sclerosis, and widespread white‐matter thorny shaped astrocytes with some predilection for frontal and limbic regions. This form of aging‐related tau astrogliopathy (ARTAG) is known for correlating with regional brain functional deficits, ²⁹ , ³⁰ and its presence in frontal and limbic regions was thought to explain some of the clinical features that led to a bvFTD syndrome. This case would not meet the new clinical criteria for LATE. ³¹

Both cases of sporadic FTLD‐TDP type A had intermediate ADNC. One had a PPA syndrome and maintained a Clinical Dementia Rating (CDR) global score of 1 for memory and orientation toward the end of his life. Family history was non‐contributory. Neocortical TDP‐43 proteinopathy was moderate. The other case had an amnestic syndrome that included forgetfulness, difficulties finding objects, diminished attention, and difficulties falling asleep, and over time, mild disinhibition, echolalia, vocalizations, and parkinsonism emerged. Neuropathological examination showed severe frontotemporal and parietal atrophy, worse on the right side, and bilateral, focal hippocampal sclerosis with selective involvement of the subiculum. In addition to intermediate ADNC, there was evidence of severe neocortical and limbic TDP‐43 proteinopathy.

3.4. Blinded assessment of MFG in LATE‐NC Stage 3+AD and FTLD‐TDP type A cases

The MFG was rated for type A (N = 39) and LATE‐NC stage 3+AD (N = 1) cases (Figure 1A‐D) by three blinded raters to assess whether examination of this one region commonly included by brain banks was sufficient to differentiate the two neuropathological diagnoses. Although the cohort included only one LATE‐NC stage 3 case, the raters were not informed of group sizes or diagnostic composition. The range in agreement with original diagnosis was 80% to 95% depending on the rater. The inter‐rater agreement was fair overall (κ = 0.391, p < 0.001), with eight cases (20%) for which the estimated diagnosis was not unanimous. Some overlap in semi‐quantitative frequency of morphological features was present (Table 4). Stratifying by burden of microvacuolation showed that for the six cases with low scores (mean < 2), only two were unanimously categorized (i.e., 67% with disagreement). In contrast, the 34 cases with a higher burden of microvacuolation (mean ≥ 2), 30 were unanimous (i.e., 12% with disagreement). In a similar analysis with stratification by frequency of neuronal cytoplasmic inclusions of TDP‐43, the 20 cases with less burden had 13 unanimously categorized cases (i.e., 35% with disagreement), whereas 19 of the 20 cases with higher scores were unanimously categorized (i.e., 5% with disagreement). Of note, all three blinded raters incorrectly categorized one specific case of FTLD‐TDP type A+AD as LATE‐NC stage 3 (Figure 1E‐H), and this case with a GRN mutation had a low burden of both microvacuolation and neuronal cytoplasmic inclusions of TDP‐43 in the MFG and a higher burden in other regions (e.g., superior, middle, and inferior temporal gyri, amygdala, entorhinal cortex, subiculum, subgenual cingulate cortex, and ventral striatum). TDP‐43 inclusions in this case included compact/round or curvilinear/crescentic cytoplasmic inclusions most prominent in small superficial layer neurons, small numbers of neuronal lentiform nuclear inclusions, and small numbers of oligodendroglial inclusions. Indeed, five of the six cases with lower burden of microvacuolation in the MFG were genetic (Table 4), and all six cases had a higher burden elsewhere (principally ventral striatum and more ventral neocortical areas). FTLD‐TDP type A cases with relatively little disease in the MFG are difficult to diagnose when assessment is restricted to that region, making them liable for misdiagnosis as LATE‐NC stage 3.

FIGURE 1.

Neuropathological hallmarks of FTLD‐TDP type A and LATE‐NC in MFG. Photomicrographs of the MFG from a FTLD‐TDP type A case with a GRN mutation stained for hematoxylin and eosin (H&E, A) and TDP‐43 (B). Normal staining for TDP‐43 localizes to the nucleus. Abnormal TDP‐43 is found in cytoplasmic inclusions (closed arrowheads) or in neurites (open arrowheads). Similar photomicrographs of the MFG from the LATE‐NC Stage 3 case are shown with H&E (C) and TDP‐43 (D) stains. Case 4 with FTLD‐TDP type A, which all three raters misclassified as LATE‐NC Stage 3 by MFG evaluation (E), is shown with regions of greater TDP‐43 burden, including subgenual cingulate (F), entorhinal cortex (G), and superior temporal gyrus (H). Panels A–D were imaged at 20× with scale bars at 50 µm, and their insets were imaged at 40× with scale bars at 10 µm. Panels E–H were imaged at 40× with scale bars at 10 µm.

TABLE 4.

Blinded ratings of MFG in FTLD‐TDP type A and LATE‐NC Stage 3.

						Average blinded middle frontal gyrus semi‐quantitative ratings from H&E and TDP‐43
Case no.	Sex	Age at death (years)	Disease duration (years)	Genetics (known autosomal dominant)	ADNC ABC score	Vacuolation	Gliosis	Neuronal loss	Neuronal intranuclear inclusions	Neuronal cytoplasmic inclusions	Dystrophic neurites	Glial cytoplasmic inclusions	Gray matter threads	White matter threads
LATE‐NC Stage 3 ± AD
1	F	86	14	Sporadic	A3 B3 C3	2.0	1.7	0.0	0.0	1.0	0.3	1.0	1.0	0.3
FTLD‐TDP Type A ± AD
2	F	82	7	Sporadic	A3 B2 C2	2.7	2.7	1.3	0.7	2.0	0.0	2.0	3.0	1.0
3	M	83	8	Sporadic	A2 B2 C2	2.3	2.3	0.3	1.0	1.3	0.0	1.0	1.7	1.0
4 b	M	72	10	GRN	A3 B3 C3	1.7	1.7	0.0	0.0	1.0	0.0	0.7	1.0	0.3
5	M	74	5	GRN	A3 B3 C2	2.7	2.7	1.0	1.0	2.7	0.0	2.0	3.0	1.3
6	M	67	13	C9ORF72	A1 B2 C2	2.7	2.0	0.7	0.3	1.0	0.0	1.0	1.3	0.7
AVERAGE (SD)						2.4 (0.4)	2.3 (0.4)	0.7 (0.5)	0.6 (0.4)	1.6 (0.7)	0 (0)	1.3 (0.6)	2.0 (0.9)	0.9 (0.4)
FTLD‐TDP type A
7 a	F	69	10	C9ORF72	A unk B2 C0	1.0	1.3	0.3	0.3	1.3	0.0	1.0	2.3	1.3
8	F	73	10	GRN	A0 B1 C0	2.7	2.7	1.0	0.7	3.0	0.0	2.3	2.7	1.7
9 a	M	64	6	C9ORF72	A0 B1 C0	1.7	1.7	0.3	0.0	2.0	0.0	1.7	2.0	1.0
10	M	72	10	Sporadic	A0 B0 C0	3.0	2.7	1.3	1.0	2.3	0.0	2.7	3.0	1.7
11	M	70	8	C9ORF72	A1 B2 C1	3.0	2.3	1.7	0.3	2.7	0.0	2.3	2.7	2.0
12	F	66	6	GRN	A1 B0 C1	2.7	3.0	2.0	0.3	2.7	0.0	2.7	3.0	1.7
13 a	M	76	9	Unknown	A0 B1 C0	2.0	1.0	0.0	0.7	1.7	0.0	1.7	1.3	1.0
14	M	68	11	GRN	A0 B0 C0	3.0	3.0	3.0	0.3	1.3	0.0	1.7	1.7	1.0
15	M	64	5	GRN	A0 B1 C0	2.3	2.3	1.3	1.3	2.7	0.3	2.3	3.0	1.3
16 a	M	63	5	Sporadic	A0 B0 C0	1.7	1.3	0.0	0.7	1.7	0.0	1.7	2.0	0.7
17	M	70	6	Sporadic	A0 B1 C0	2.0	2.0	0.3	0.3	1.3	0.0	1.7	2.0	1.0
18 a	F	79	8	Unknown	A3 B0 C3	2.3	1.7	0.3	0.0	1.3	0.0	1.7	2.3	1.0
19 a	F	74	8	Unknown	A1 B1 C unk	2.7	2.0	1.0	0.3	1.3	0.0	1.3	2.0	1.0
20	F	59	12	GRN	A1 B1 C1	3.0	3.0	2.0	0.0	1.3	0.0	0.7	1.7	1.0
21	F	61	7	GRN	A0 B1 C0	2.7	3.0	1.0	1.0	2.7	0.0	2.0	2.3	1.3
22	F	70	10	GRN	A1 B1 C0	2.0	2.0	0.3	1.0	2.3	0.0	2.3	2.7	1.3
23	M	68	5	Unknown	A0 B2 C0	2.0	1.3	0.0	0.7	2.7	0.0	2.0	2.3	1.0
24	F	56	5	GRN	A1 B1 C0	2.3	3.0	1.0	0.7	2.3	0.0	2.0	2.7	1.7
25	F	78	8	GRN	A1 B0 C2	2.3	2.0	0.7	0.0	1.3	0.0	1.3	2.0	1.3
26	F	66	5	GRN	A0 B0 C0	3.0	2.7	1.7	1.0	2.7	0.0	2.7	3.0	1.7
27	F	73	8	GRN	A1 B1 C0	2.7	2.7	0.7	1.0	2.7	0.0	2.0	3.0	1.0
28	F	64	8	GRN	A1 B1 C0	2.3	2.0	0.3	0.7	1.7	0.0	1.7	2.3	2.0
29 a	F	78	7	GRN	A0 B1 C0	2.0	2.0	0.3	0.3	1.3	0.0	1.3	2.0	1.0
30	F	84	21	GRN	A1 B0 C0	3.0	3.0	2.3	0.0	1.7	0.0	1.3	1.7	1.0
31	F	68	8	C9ORF72	A1 B2 C0	2.0	1.3	0.3	0.7	1.7	0.0	1.7	2.0	1.0
32	F	63	6	GRN	A3 B1 C2	3.0	3.0	2.3	1.0	2.3	0.3	2.7	3.0	1.3
33	F	74	7	GRN	A1 B1 C1	1.3	0.7	0.0	0.3	2.7	0.0	2.3	3.0	1.3
34	M	69	6	GRN	A1 B1 C0	3.0	2.3	1.3	1.0	1.7	0.0	2.0	2.7	1.3
35 a	F	84	3	GRN	A1 B2 C1	1.7	0.3	0.0	0.7	1.7	0.0	2.3	1.7	1.0
36	M	57	5	GRN	A1 B2 C0	3.0	2.3	1.0	1.0	3.0	0.0	2.7	3.0	2.0
37	M	65	4	GRN	A1 B1 C0	2.0	1.7	0.3	1.3	2.0	0.3	2.0	2.3	1.3
38	F	70	7	C9ORF72	A1 B1 C0	2.0	1.3	0.7	0.7	1.3	0.0	1.0	1.7	0.3
39	M	70	10	GRN	A0 B2 C0	2.3	2.0	0.0	0.7	2.3	0.3	2.0	3.0	2.0
40	F	52	5	GRN	A0 B1 C0	2.3	2.0	0.7	1.0	2.7	0.0	2.7	3.0	2.0
AVERAGE (SD)						2.4 (0.5)	2.1 (0.7)	0.9 (0.8)	0.6 (0.4)	2.0 (0.6)	0 (0.1)	1.9 (0.5)	2.4 (0.5)	1.3 (0.4)

Note: ADNC ABC scores in two cases were unknown (unk) due to incomplete assessment by an outside, initial evaluation, as noted above.

^aCases where at least one rater classified the case as LATE‐NC based on the blinded MFG assessment.

^bCase where all three raters agreed on the opposite diagnosis from the original assessment of the entire case.

4. DISCUSSION

The neuropathological entities of FTLD‐TDP and LATE‐NC share some histopathological features, ³² and their defining boundaries are still being drawn and refined. ³ In the present study from the UCSF NDBB, which specializes in FTLD, 190 cases with TDP‐43 proteinopathy were analyzed, and 78% were classified as FTLD‐TDP and 22% as LATE‐NC+AD. Although previous studies compared them, most had few FTLD‐TDP cases, especially type A. We discuss these previous studies and incorporate the present study below to determine whether a unified view of these diagnoses can be rendered.

4.1. Arguments favoring a single disease spectrum

Early histological observations raised concerns that FTLD‐TDP and LATE‐NC might reside on a spectrum of the same disease. The neuropathological characteristics of type A and LATE‐NC can, in some cases, bear a striking resemblance with respect to TDP‐43 inclusions and, to some degree, the spatial topology. Tomé et al. recently showed that the quantification of dystrophic neurites and neuronal cytoplasmic inclusions is similar in the amygdala and hippocampus between pure LATE‐NC and FTLD‐TDP type A cases; further, these two groups often had a mesh‐like neuritic TDP‐43 pattern in the hippocampus, from CA 1 and 2 to the subiculum. ³³ They also showed similar rates of nuclear clearing of TDP‐43 in these same structures. ³³ FTLD‐TDP typically has a younger symptom onset, while LATE‐NC is associated with older age, making age and comorbid disease (especially AD) difficult to match in studies. Buciec et al. ¹¹ compared genetic FTLD‐TDP cases (any type) with older ages of onset (N = 15) to age‐matched cases of LATE‐NC with or without coexistent AD (N = 20), and they found a limbic‐predominant distribution of TDP‐43 proteinopathy across both groups that made differentiating them difficult. In comparisons across many brain regions, only through quantification was a greater TDP‐43 burden detected in FTLD‐TDP over LATE‐NC, specifically in the entorhinal cortex, CA1, and the subiculum; however, the MFG burden was similar. Further, genetic FTLD‐TDP cases with older onset often clinically resembled LATE‐NC, with AD‐type dementia instead of FTD syndromes, though this observation was based on very few cases and requires replication.

Other studies reported shared genetic risk. Autosomal‐dominant mutations in GRN result in FTLD‐TDP type A, whereas the common GRN rs5848 variant, which lowers circulating progranulin protein levels, elevates LATE‐NC risk. ³⁴ , ³⁵ , ³⁶ In our cohort, the TMEM106B rs1990622 A/A genotype, which is associated with altered phenotype and pathological severity of FTLD/ALS‐TDP cases, ³⁷ , ³⁸ , ³⁹ , ⁴⁰ , ⁴¹ was observed at a similar frequency in FTLD‐TDP and LATE‐NC. In a prior study, our cohort showed similar TMEM106B minor allele frequency among all TDP‐43 cases, except for LATE‐NC, and when compared to a population‐based autopsy cohort. ⁴² Other studies reported an association between this TMEM106B genotype and LATE‐NC. ³⁵ , ³⁶ , ⁴³ It may be that the TMEM106B genotype increases risk for TDP‐43 proteinopathy in general, though confirming this will require a large genetic study that includes all forms of TDP‐43 pathology. A further complication in the genetics literature is pleiotropy. For example, GRN rs5848 has been associated with increased risk for Parkinson's disease and AD, ⁴⁴ and TMEM106B may also influence the clinical course of these disorders. ⁴⁵ , ⁴⁶ Thus, some genetic variants may confer a broader susceptibility to TDP‐43–related or other neurodegenerative proteinopathies, rather than specifically linking FTLD‐TDP and LATE‐NC within a single disease spectrum.

4.2. Arguments favoring separate entities

Several studies sought to leverage larger datasets to differentiate FTLD‐TDP and LATE‐NC. In an unsupervised clustering analysis of 495 cases from the NACC autopsy dataset, Katsumata et al. ¹⁶ identified four TDP‐43 proteinopathy subtypes based on neuropathological data and age at death, and they compared their clinical and genetic characteristics. By inclusion criteria, all cases had evidence of TDP‐43 proteinopathy in the amygdala, hippocampus, or MFG. FTLD‐TDP (any type) was reported in 107 cases (22%) from this cohort. The four clusters were divided by age at death as greater or less than 85 years and the relative proportion of ADNC and neocortical TDP‐43 (specifically, low or intermediate ADNC with greater neocortical TDP‐43, or vice versa), representing well some of the most common clinicopathological profiles seen with age‐related neurodegeneration. A lower ADNC was found in the FTLD‐TDP and LATE‐NC groups, and the early‐ and late‐onset AD cases comprised the two clusters with higher ADNC. Of note, only 30% of the LATE‐NC cluster included neocortical TDP‐43 proteinopathy, and only 25% of cases in the early‐onset AD‐corresponding cluster had neocortical involvement of TDP‐43. Vacuolation, astrogliosis, and neuronal loss were not assessed. This study had several strengths, including the fact that many cases had cognitive domains tested during longitudinal assessments and had state‐of‐the‐art neuropathological evaluations at high‐quality academic research centers. Unfortunately, this referral bias limits the generalizability of the findings to the broader population of community‐dwelling people, who are less likely to present to academic centers. Further, for scalable, pathology‐based criteria to differentiate FTLD‐TDP from LATE‐NC, sharper tools than the currently available parameters from the NACC dataset will be needed.

In a different analysis of the NACC dataset, Teylan et al. ¹⁷ interrogated differences in symptomatic presentation, cognitive performance, and other characteristics from the Uniform Data Set (UDS) for cases with LATE‐NC (N = 265) or FTLD‐TDP (any type; N = 92), and they found that participants with LATE‐NC were older at death (83.6 vs 72.3 years), were more likely to carry the APOE ε4 allele, had higher ADNC, and had better cognition and lower CDR global score at death. In a follow‐up analysis, participants with LATE‐NC and final CDR global score ≥2 before death were more likely to have visuospatial impairment, delusions, or visual hallucinations, and these differences held after sensitivity analyses that mitigated older age at death, LATE‐NC stage 3, or high burden of ADNC. Due to the debate as to whether FTLD‐TDP and LATE‐NC are the same disease with different manifestations depending on age, which parallels the faster progression of early‐onset AD, the researchers then focused their analysis on FTLD‐TDP and LATE‐NC cases with a relatively younger age at death of less than 80 years. Most differences persisted (eight of 11 cognitive or behavioral symptoms), though the difference in CDR global scores did not. The large sample with uniform data collection was a strength, though some flaws were inherent in the sample and the study design. First, FTLD‐TDP was analyzed without a breakdown by its subtypes, which are not included in the NACC dataset, and ideally LATE‐NC would be compared directly to type A. Second, the sensitivity analyses reduced the samples of certain groups, notably the LATE‐NC participants under 80 years of age at death (N = 79) versus FTLD‐TDP under 80 years (N = 73). Third, vacuolation, astrogliosis, and neuronal loss, which can aid in identifying FTLD‐TDP cases using routine workflows across many brain banks because by classical definition these are FTLD features, ⁴⁷ were not assessed. Lastly, the high ADNC associated with LATE‐NC likely drove a substantial portion of the symptoms and may have masked other, more subtle clinical effects due to LATE‐NC.

Addressing FTLD‐TDP subtype heterogeneity and contrasting FTLD‐TDP subtypes with LATE‐NC, Robinson et al. ¹⁸ used an unsupervised clustering method of TDP‐43 morphological features assessed by blinded raters across four brain regions in a combined cohort of 41 FTLD‐TDP (N = 27 of TDP type A or B, which were not separated) and 44 LATE‐NC (N = 33 at stage 3) participants from two academic centers, and they obtained excellent separation between FTLD‐TDP types C and E from LATE‐NC and other FTLD‐TDP subgroups. Clustering analysis performed poorly in separating the combined FTLD‐TDP types A/B from LATE‐NC (with only 83% accuracy). Whereas both FTLD‐TDP and LATE‐NC groups had similar burdens of TDP‐43 proteinopathy in the amygdala, neocortical involvement typically was more severe in FTLD‐TDP; specifically, the neuronal cytoplasmic inclusions and dystrophic neurites were more severe in the superior temporal and middle frontal cortices as well as in the anterior cingulate. Despite the strength of this study's high proportion of LATE‐NC stage 3 cases, the analysis includes all FTLD‐TDP types combined, rather than comparing LATE‐NC to FTLD‐TDP types A or B.

4.3. Experience of a brain bank specializing in FTLD‐TDP

This study leveraged detailed neuropathological, clinical, and genetic data to examine subtle differences between FTLD‐TDP and LATE‐NC+AD. The analysis was driven conceptually by the histopathological similarities between FTLD‐TDP type A and LATE‐NC. As such, the FTLD‐TDP subtypes were described separately, including type U cases that more often show mixed A and B features or display too sparse pathology to allow subtyping. In our cohort, as expected, FTLD‐TDP type A had a younger onset, shorter disease duration, and younger age at death than LATE‐NC+AD. FTLD‐TDP+AD cases had a more intermediate age at death compared with the other two groups. Moreover, all LATE‐NC+AD cases were sporadic, whereas FTLD‐TDP type A often had an autosomal‐dominant genetic cause. In addition, the APOE ε4 allele frequency was much higher in LATE‐NC+AD than in FTLD‐TDP type A without AD. Finally, amnestic, “AD‐type” clinical diagnoses were given very often to cases with LATE‐NC+AD, and FTD syndromes typically were used for those who were type A at autopsy. As all LATE‐NC cases in our collection and in most of the literature had intermediate to high ADNC, we performed a qualitative subanalysis of the seven FTLD‐TDP cases with type A inclusions who also had intermediate to high ADNC. Of note, three of them received a clinical diagnosis of AD‐type dementia. As there are no established neuropathological criteria to distinguish LATE‐NC from FTLD‐TDP type A, one could argue that these cases should be classified as LATE‐NC stage 3. However, two of them were familial (GRN and C9ORF72). Not all brain banks have access to genetic results or family history. Thus, an amnestic presentation does not preclude a FTLD‐TDP diagnosis. ⁴⁸ , ⁴⁹ , ⁵⁰ , ⁵¹

Next, given that the UCSF NDBB has a strong interest in TDP‐43 pathology, our workup for TDP‐43 tends to be more extensive than in most brain banks, giving us a broader understanding of the extent of TDP‐43 pathology and severity of neurodegeneration in our cases. Along these lines, as our cohort has a larger subset (N = 39) of FTLD‐TDP cases with type A inclusions than previous reports, we initially aimed to perform a head‐to‐head comparison between FTLD‐TDP type A and LATE‐NC stage 3 (+AD). However, only one of our 42 LATE‐NC cases reached stage 3, impeding such an analysis and likely reflecting the referral bias for our center toward younger‐onset, atypical presentations of AD and FTD. Nevertheless, we pivoted and leveraged the strengths of our collection to interrogate whether the severity of MFG neurodegeneration plus TDP‐43 pathology was enough to distinguish LATE‐NC stage 3+AD from FTLD‐TDP type A by examining blinded digitized MFG slides immunostained for TDP‐43. In addition, given neurodegeneration of superficial cortical layers (microvacuolation, neuronal loss, and astrogliosis) is an important, defining criterion for diagnosis of FTLD, whereas LATE‐NC tends to show milder neurodegeneration patterns in the neocortex, we also included parallel slides stained for H&E in the analysis. One case originally classified as FTLD‐TDP type A was misclassified as LATE‐NC by all raters and another eight by at least one of the three raters. The case misclassified by all raters had little MFG pathology and carried a GRN mutation, and the severity of neurodegeneration of the other eight was relatively low for the MFG. Our results suggest that ruling out FTLD‐TDP in cases with TDP‐43 pathology in the MFG requires investigation of other neocortical regions; otherwise, a misdiagnosis of LATE‐NC stage 3 may be made. As there is only one LATE‐NC stage 3 case, these results on the MFG evaluated in isolation should be considered exploratory. The literature suggests the inferior frontal gyrus (opercular region, which corresponds to Broca's area) or orbitofrontal gyri, which often are involved early in FTLD‐TDP when presenting as bvFTD ⁵² and are involved late in LATE‐NC, ⁵³ may be candidate regions. Our results also suggest that assessing genetic and clinical data can help diagnostically, but they are not always available in non‐ultra‐specialized brain banks. Lastly, our findings extend prior work ¹⁹ by showing that examining the MFG is necessary but often insufficient in diagnosing some FTLD‐TDP type A cases.

4.4. Conclusion and future steps

Although most of the literature supports FTLD‐TDP and LATE‐NC as separate entities, especially FTLD‐TDP types B, C, D, and E, whether FTLD‐TDP type A and LATE‐NC reside on the same spectrum remains debatable. Even machine learning algorithms have difficulty distinguishing late‐stage LATE‐NC from early‐stage FTLD‐TDP. ⁸ One could argue that both represent a convergence of pathogenetic mechanisms, a model used to explain AD heterogeneity. As a path forward, the best analogy may be the comparison of FTLD‐TDP type B to ALS‐TDP (Table 5). They share the same genetic risk, the same protein fold by cryo‐electron microscopy, ⁵⁴ the same pathomorphology, and related populations of selectively vulnerable neurons, all features that could argue for viewing them as a single disease. Still, even while their affected regions and clinical features may converge as they progress, they likely differ in their neuroanatomic origins and clinicopathological progression, making it useful to maintain them as distinct entities.

TABLE 5.

Analogous comparisons between neuropathological entities.

	Comparison 1		Comparison 2
	FTLD‐TDP type B	ALS‐TDP	FTLD‐TDP type A	LATE‐NC
Genetic loci examples	C9ORF72, TARDBP, TBK1	C9ORF72, TARDBP, TBK1	GRN, C9ORF72, TMEM106B	APOE, GRN (variant not autosomal dominant), TMEM106B
Protein fold	Double spiral	Double spiral	Chevron (V shape)	Unknown
Pathomorphology	Loosely defined: neuronal intracytoplasmic inclusions appear speckled, globular, or skein‐like. Threads are not typically present.	Loosely defined: neuronal intracytoplasmic inclusions appear speckled, globular, or skein‐like. Threads are not typically present.	Loosely defined: neuronal intracytoplasmic inclusions appear comma‐shaped or globular, and intranuclear inclusions may appear as cat eye shape. These occur in among small threads and oligodendroglial inclusions.	Relatively undefined: neuronal intracytoplasmic inclusions and threads are described.
Selectively vulnerable neurons	Von Economo and other layer 5 extratelencephalic neurons, small upper layer pyramidal neurons	Upper and lower motor neurons	Von Economo and other layer 5 extratelencephalic neurons, small upper layer pyramidal neurons	Neurons in ventral amygdala and CA1/subiculum and dentate gyrus of hippocampus.
Neuropathological staging system and neuroanatomy	No staging system. Predominantly fronto‐insulo‐temporal atrophy.	No widely adopted staging system. Corticospinal tract, motor cranial nerves, and anterior horns of spinal cord.	No staging system. Predominantly fronto‐insulo‐temporal atrophy	Three‐stage system of amygdala, hippocampus, and MFG.
Clinical syndromes and progression	FTD spectrum, sometimes includes evidence for motor neuron disease.	ALS, sometimes includes behavioral or cognitive features from FTD spectrum.	FTD spectrum	Primarily amnestic and recently codified in LATE criteria. Limited clinical description of advanced (Stage 3) cases.
Preliminary conclusion	Likely represent divergent manifestations of a single disease but are useful to keep as separate entities.		Debated but currently useful to keep as distinct entities.

A key question for the convergence hypothesis for FTLD‐TDP type A and LATE‐NC is what leads to their similar‐appearing inclusions. The genetic features of FTLD‐TDP and LATE‐NC are distinct. Cryo‐electron microscopy analysis for misfolded TDP‐43 in LATE‐NC has not been published. Finally, there is some pleomorphism in LATE‐NC cases, with not all adopting classical FTLD‐TDP type A features. Therefore, unification on grounds of shared protein structure cannot be achieved until a LATE‐NC aggregate structure is solved and type‐specific antibodies are developed. At present, keeping these as separate diagnostic terms remains justified for clinical and research purposes, as is done for FTLD‐TDP type B and ALS‐TDP. Ultimately, improving biomarker, neuropathological, and genetic criteria to differentiate FTLD‐TDP and LATE‐NC is necessary for future work to explore TDP‐43 proteinopathy and the heterogeneity in its manifestations.

Finally, our results suggest that, at least for now, in cases with TDP‐43 pathology in the MFG, the TDP‐43 proteinopathy needs to be examined for other neocortical areas to help differentiate FTLD‐TDP type A from LATE‐NC stage 3.

CONFLICT OF INTEREST STATEMENT

None directly regarding this manuscript for all authors. For the complete list of potential conflicts of interest, please refer to the International Committee of Medical Journal Editors (ICJME) document. Author disclosures are available in the Supporting Information.

CONSENT STATEMENT

Informed consent was obtained from all human participants for participation in this research and for brain donation.

Supporting information

Supporting Information: dad270411‐sup‐0001‐ICMJE.pdf