. Limbic-predominant age-related TDP-43 encephalopathy differs from frontotemporal lobar degeneration. Brain. 2020 Sep 1;143(9):2844-2857. PubMed.


Please login to recommend the paper.


  1. This is an interesting and carefully conducted analysis comparing TDP-43 pathology in FTLD and LATE that describes differences in disease phenomenology assessed at autopsy. The study will be a good starting point to apply biochemical and molecular biological techniques to better understand the mechanistic relationships between these phenomenologically distinct TDP-43 proteinopathies. Comparing these diseases may have important therapeutic implications since TDP-43-directed therapies being developed for FTLD could eventually be useful in LATE +/- AD if the pathogenic mechanisms are related.

  2. This is an interesting study to address the question of how to separate cases with TDP-43 pathology in the context of other disease and/or aging (i.e., LATE-NC) from primary TDP-43 proteinopathies, particularly FTLD-TDP, based on neuropathological features. The authors did a great job evaluating the density of distinct TDP-43 positive inclusions semiquantitatively in preselected cases. This extended and confirmed previous observations, namely that FTLD-TDP cases usually show more severe pathology in the frontal and cingulate gyrus than those classified as LATE-NC.

    However, in my opinion, translation of this finding (that mild pathology in frontal areas indicates LATE-NC, while moderate to severe pathology in frontal equates to FTLD-TDP) into the routine diagnosis of individual cases still faces some challenges. First, semiquantitative grading is known to be highly subjective and expected to vary considerably from lab to lab (as seen for other neuropathological criteria using semiquantitative scores). Second, as illustrated by the authors and reported previously, there are cases considered to represent LATE-NC with moderate to severe TDP-43 pathology in regions overlapping those affected in FTLD-TDP cases, and vice versa. It will be very interesting to see the validity/reliability of suggested criteria in a multicenter study.

    Finally, it will be very important to see how suggested criteria for labeling a case as LATE-NC will work or have to be adapted in ALS, ALSci, and ALSbi cases which, if presenting with cortical TDP-43 pathology, often have milder frontal TDP-43 pathology than FTLD-TDP cases.

  3. This is the type of work necessary to determine if LATE-NC is a distinct pathological entity beyond AD with TDP-immunoreactive co-pathology. These investigators have shown that there are group-level differences in several TDP-43-related neuropathologic features between FTLD-TDP and two subtypes of LATE-NC. The Penn LATE-NC cases mostly had primary AD pathology (28 of 30 volunteers), while the Kentucky LATE-NC cases had little to no AD pathology, it being limited to the hippocampus in nine of the 14 cases.

    The study provides evidence that 1) there are group-level differences in several TDP-immunoreactive (TDP-ir) features, with some significant overlap; 2) neuropathologists can reliably distinguish between FTLD-TDP and non-FTLD TDP-ir lesions.

    At the same time, it again shows that—even in centers with very large numbers of neuropathologic examinations—LATE-NC in the absence of AD is very rare, and many of these AD-negative cases could have hippocampal sclerosis dementia (a rare, yet known, entity). It would be useful to know whether the 11 cases with hippocampal-only TDP-ir lesions were previously diagnosed with hippocampal sclerosis. The remaining five cases from Kentucky are also of interest, and it would be informative to know if these cases had a collection of limited regional pathology as reported previously (Petersen et al., 2006). Finally, as semiquantitative scoring overlooks the great variability in lesional density within TDP-43 neuropathological score category 3 during the human diagnostic process, a more objective/unbiased classification algorithm should be used to test whether semiquantitative TDP-ir features alone can distinguish between FTLD-TDP and LATE-NC.

    There remain significant questions and interest in how TDP-ir lesions mechanistically contribute to cognitive decline in AD (are they additive, synergistic, serve as a biomarker, or simply bystander) based on published data (e.g., chimeric Aβ-TDP oligomers). Until we have better understanding of the biological differences among TDP-ir lesions—beyond morphological similarities—in AD with TDP-ir lesions, in hippocampal sclerosis, and in the rare dementia cases with limited pathologic changes, LATE-NC should be reserved for those cases without moderate to high likelihood of AD.


    . Neuropathologic features of amnestic mild cognitive impairment. Arch Neurol. 2006 May;63(5):665-72. PubMed.

  4. This study addresses the important issue of the relationship between FTLD-TDP and what the authors refer to as LATE-NC. Unfortunately, the study design limits the value of the results. Because the cases were selected on the basis of both clinical features and pathology, all that has been shown is that patients with clinical FTD have more abundant neocortical TDP-43 pathology than those who do not have FTD. This is not terribly surprising or informative. In addition, the study did not identify any qualitative differences that distinguish the two groups, but propose a “criteria” that is based simply on the semiquantitative burden of total TDP-43 pathology.

    So, this doesn’t really answer the question of whether FTLD-TDP and LATE-NC are fundamentally different or a continuum. We still don’t know if LATE-NC can progress to cause FTD, because then it would simply be diagnosed as FTLD-TDP. And using the criteria that have been suggested, a case with the C9ORF72 mutation who died of ALS in the early stages of FTD would be classified as having LATE-NC rather than mild FTLD-TDP.

  5. The comments above raise important issues related to an understudied but common disease(s). I first offer individual responses, then some overarching points.

    I agree with Dr. Boxer that therapeutics for one subtype of TDP-43 proteinopathy may work for another. TDP-43-related biomarkers might also have cross-disease utility. There are glaring and perhaps overlapping needs affecting different groups of patients!

    Dr. Neumann’s points about the challenges of semiquantitative pathologic methods are well-taken. It would be great to have access to dichotomous or qualitative parameters to differentiate LATE-NC and FTLD-TDP. However, I note that only about 20 percent of LATE-NC cases have any frontal cortical TDP-43 proteinopathy, and a fair number of FTLD-TDP cases are diagnosable as subtypes C-E. For these cases, there should be no overlap at all, even without reliance on semi-quantitative parameters. As for the other cases, there were large differences in frontal cortical TDP-43 proteinopathy in FTLD-TDP cases versus LATE-NC cases, but more work is needed there.

    With regard to there being some cases (fewer than 2 percent of TDP-43+ cases overall in the U.Penn. and U. Kentucky biobanks) that would have diagnostic overlap according to the blinded raters in the present study: Some may say 98 percent reliability for a classification scheme is bad. However, reliability is questionable for other neuropathologic diagnostic classifications, e.g., FTLD subtypes alpha versus beta, PSP versus CBD, etc. I think the simple LATE/FTLD method we described in Robinson et al. will probably work as well, if not better, than most.

    The point that this diagnostic criterion may have to be adapted for ALS/ALSci/ALSbi is a great one. It’s an open question (many ALS subjects have amnestic problems), and I have little experience with ALS. I’d love to see more data on that.

    I also agree with the overall points made by Dr. Hu, but would like to address two things. First, it is not true that LATE-NC in the absence of AD is very rare. It is far more common than FTLD or ALS. A large number of cases in community-based cohorts such as ours have LATE-NC but lack ADNC. We only used a subset of cases in this paper—those with the most frontal cortex TDP-43 proteinopathy; most of our cases lacking ADNC also lacked frontal cortex TDP-43 proteinopathy. (Same for ADNC+ cases.)

    This is true in other cohorts, too. In a community-based cohort, persons were categorized into (1) no pathologic diagnosis of LATE or ADNC (n=378), (2) LATE-NC without ADNC (n=91), (3) ADNC without LATE-NC (n=535), and (4) mixed ADNC with LATE-NC (n=352). In other words, in this sample, roughly 80 percent of subjects with LATE-NC had ADNC. However, in subjects without LATE-NC, almost 60 percent had ADNC (Kapasi et al., 2020). This indicates that there is probably some synergy between ADNC and LATE-NC, but there are plenty of cases with LATE-NC lacking ADNC.

    Second, I respectfully disagree that LATE-NC should be reserved for those cases without moderate to high likelihood of AD. LATE-NC is clearly associated with significantly worse clinical outcomes for those with comorbid ADNC, according to research labs from around the globe. Given that the clinical course and final outcomes are far worse for people with ADNC+LATE-NC than those with ADNC alone (Kapasi et al., 2020; Katsumata et al., 2020; Karanth et al., 2020; Nelson et al., 2010; and many others), it is a highly relevant separate diagnosis that belongs on the top line of autopsy reports.

    I appreciate Dr. Mackenzie’s healthy skepticism and agree that the study did not identify any qualitative differences that distinguish LATE-NC and FTLD-TDP. However, this commentary misses some key points.

    First, an important way to consider a disease is to include both clinical and pathological information. In our study, we established that people with the FTLD syndrome (with the clinical features factored in) have, as expected, significantly different clinical features than those of LATE-NC with advanced TDP-43 proteinopathy—none of the latter subjects had full-blown FTD clinical syndrome. And the ages of death for the groups were almost non-overlapping. Further, the individuals with the FTD clinical syndrome had substantially more total cortical TDP-43 proteinopathy than those (with or without AD) with the most frontal cortical TDP-43 pathology in two different biobanks.

    The question of whether FTLD-TDP and LATE-NC are a “continuum,” in the sense of querying whether LATE-NC cases will generally progress to FTLD-TDP, has essentially been answered by community-based cohorts who follow subjects from normal cognitive status to death. In short, FTLD-TDP and LATE-NC are not a continuum in this sense. Many elderly die with LATE-NC and very few TDP-43 proteinopathy sufferers past age 90 die with an FTD/FTLD clinical or pathological picture. It is simply not the case that LATE-NC tends to progress to FTLD-TDP. I’m not sure why. An intriguing parallel question is: Why does PART-type tauopathy not progress past Braak NFT Stage IV in the absence of Aβ plaques? FTLD-TDP and LATE-NC do, however, have some areas of overlap.

    As to whether a case with the C9ORF72 mutation who died of ALS in the early stages of FTD would be classified as having LATE-NC rather than mild FTLD-TDP, I’m ignorant about this, but interested. I'd like to see a study comparing early stages of ALS/FTLD spectrum pathology with LATE-NC. I am agnostic what those results would look like.

    An important point is that TDP-43 proteinopathy occurs in many different neurological conditions (Chornenkyy et al., 2019). The notion that “there is TDP pathology, so it’s either ALS or FTLD” is not sustainable. Given that FTLD-TDP has different age-of-onset and clinical features than LATE-NC, it seems suboptimal to render this diagnosis for persons lacking those clinical signs and symptoms ... that would only cause needless confusion for clinicians and families who read the autopsy report.

    Overall, then, I summarize our current state of knowledge as follows:

    1. Dementia clinic-based (DCB) researchers have very different experiences than those who evaluate autopsy results in community-based (CoB) cohorts. DCB cohorts are strongly enriched for FTLD and ALS cases. CoB see lots of LATE. I personally think that a lot of the controversy in this field is caused by the differing perspectives of DCB and CoB-based researchers. A strength of Robinson et al. was the combination of samples and diagnosticians from DCB and CoB centers.

    2. LATE neuropathologic changes (LATE-NC) are more than 100 times more prevalent than ALS/FTLD (Coyle-Gilchrist et al., 2016; Knopman and Roberts, 2011), and affect a different age range, but this perspective may be hard to appreciate by DCB researchers, who generally draw from large catchment areas, see so much FTLD, and also see the even rarer outliers.

    3. According to CoB cohorts, there is no evidence that LATE-NC tends to evolve to FTLD. There is plenty of evidence that many persons, aged 90-105, die with LATE-NC and not an FTLD-type disease. Around 2 percent of FTLD cases lived past age 85 (Katsumata et al., 2020). 

    4. Nonetheless, there are overlapping biological pathways, and genetic pleiotropy, related to LATE and AD neuropathologic changes (ADNC), as well as to LATE and FTLD.

    5. Approximately 80 percent of persons tend to have some ADNC—even more in DCB cohorts. Therefore, that most cases of LATE-NC (usually older than 85) have comorbid ADNC does not indicate that TDP-43 proteinopathy is always downstream of ADNC, because even if they occurred randomly, ~80 percent of persons with LATE-NC would have comorbid ADNC.

    6. As with amygdalar Lewy bodies, amygdalar TDP-43 proteinopathy is probably often a downstream feature of ADNC. But it's still not seen in every case of severe ADNC (Besser et al., 2020). 

    7. LATE-NC is associated with worse clinical features for a given severity of ADNC. This has been found by many different investigators in many different cohorts. Therefore, it is important to include LATE-NC in the autopsy diagnostic report when the pathology is present.

    8. LATE-NC preferentially affects the medial temporal structures and frontal regions, as does FTLD. LATE-NC is affected by risk genes (TMEM106B and GRN) that are implicated in other conditions, including FTLD. The pathologies of LATE-NC and FTLD-TDP (type A) overlap in some ways. Some FTLD cases are old, some LATE-NC cases are younger, so there is a need for more research at the biologic and diagnostic boundary zones.


    . Limbic-predominant age-related TDP-43 encephalopathy, ADNC pathology, and cognitive decline in aging. Neurology. 2020 Oct 6;95(14):e1951-e1962. Epub 2020 Aug 4 PubMed.

    . Distinct clinicopathologic clusters of persons with TDP-43 proteinopathy. Acta Neuropathol. 2020 Nov;140(5):659-674. Epub 2020 Aug 14 PubMed.

    . Prevalence and Clinical Phenotype of Quadruple Misfolded Proteins in Older Adults. JAMA Neurol. 2020 Jun 22; PubMed.

    . Modeling the association between 43 different clinical and pathological variables and the severity of cognitive impairment in a large autopsy cohort of elderly persons. Brain Pathol. 2010 Jan;20(1):66-79. PubMed.

    . Tau and TDP-43 proteinopathies: kindred pathologic cascades and genetic pleiotropy. Lab Invest. 2019 Jul;99(7):993-1007. Epub 2019 Feb 11 PubMed.

    . Prevalence, characteristics, and survival of frontotemporal lobar degeneration syndromes. Neurology. 2016 May 3;86(18):1736-43. Epub 2016 Apr 1 PubMed.

    . Estimating the number of persons with frontotemporal lobar degeneration in the US population. J Mol Neurosci. 2011 Nov;45(3):330-5. PubMed.

    . Limbic Predominant Age-Related TDP-43 Encephalopathy (LATE): Clinical and Neuropathological Associations. J Neuropathol Exp Neurol. 2020 Mar 1;79(3):305-313. PubMed.

  6. I thank Dr. Nelson for responding. Often we do not hear back from the senior authors on these posts, so the effort to continue this debate is encouraging.

    I will only comment on the issue of prevalence. Much of the data on this has come from the impressive studies carried out by Drs. Bennett and Schneider at the Rush ADRC. However, a central problem with the statistical analysis is that ADNC and LATE-NC are measured on different scales. ADNC has been dichotomized (1=moderate/high likelihood AD; 0=no or low likelihood AD), which means many "ADNC negative" cases in this analysis—while not fulfilling AD pathologic diagnosis—are not free of ADNC. On the other hand, LATE-NC is at least ordinal, and I suspect many of them can be considered FTLD-negative. Thus, the two pathologic changes are measured on different scales and do not share metric invariance. There are two approaches to resolve this controversy:

    1. TDP-ir lesions in all cases can be dichotomized, not by whether LATE-NC is present, but by whether there are sufficient TDP-ir lesions to be considered FTLD-TDP (based on scaling used in ADNC dichotomization). I believe any LATE-NC cases identified through this approach would fall under FTLD-TDP.

    2. Alternatively, ADNC can be maintained as ordinal values according to plaque and tangle distribution (one ordinal score for A, and one ordinal score for B). This will provide more accurate information on cases without sufficient ADNC for diagnostic certainty. Because I believe Dr. Nelson would agree that FTLD-TDP cases have much greater TDP-ir pathology than LATE-NC cases, power will not be significantly reduced in this approach as 0-3 LATE-NC scale does not cover as wide a topographical range as 0-3 A or B scores.

Make a Comment

To make a comment you must login or register.