Abstract

Open in a separate window

TDP-43 proteinopathy may lower the threshold for clinical expression of Alzheimer-type dementia. James et al. investigate the relationship between TDP-43 and Alzheimer’s dementia in a large cohort of community-dwelling older adults. Mixed Alzheimer/TDP-43 pathology is associated with increased likelihood of clinical Alzheimer-type dementia proximate to death, independent of other pathologies.

Abstract

Introduction

Hyperphosphorylated transactive response DNA-binding protein 43 (TDP-43, encoded by TARDBP) is a pathological protein associated with frontal temporal lobar degeneration (FTLD) and amyotrophic lateral sclerosis (ALS) (Neumann et al., 2006). It has recently been shown to be common in older brains (Geser et al., 2010; Arnold et al., 2013), is reported to be present in at least one in five and up to two in three brains with pathologically confirmed Alzheimer’s disease (Amador-Ortiz et al., 2007; Uryu et al., 2008; Tremblay et al., 2011), and may progress through the brain in a stereotypical manner in patients with Alzheimer’s disease pathology (Hu et al., 2008; Josephs et al., 2014a, 2016). In subjects with pathologically confirmed Alzheimer’s disease, TDP-34 was associated with greater memory loss and hippocampal atrophy (Josephs et al., 2008, 2014b). In other studies in patients both with and without pathological Alzheimer’s disease, TDP-43 was associated with greater cognitive impairment and dementia (Nelson et al., 2010; Robinson et al., 2011; Wilson et al., 2013). Taken altogether, this suggests that TDP-43 may lower the threshold for the clinical expression of Alzheimer’s-type dementia in much the same way that other common comorbid pathologies have been shown to contribute to the clinical expression of Alzheimer’s disease. However, the relationship between TDP-43 and the clinical expression of Alzheimer’s-type dementia has not been systematically investigated in a large cohort of community-dwelling older subjects both with and without Alzheimer’s disease neuropathology, and with and without dementia.

We previously showed that co-occurrences of Alzheimer’s disease with other common dementia-related neuropathologies, i.e. mixed pathologies, increase the odds of developing dementia before death (Schneider et al., 2007), and that mixed pathologies account for most cases of clinical Alzheimer’s-type dementia (Schneider et al., 2009b). Thus far, the field of dementia research has primarily focused on the pathologies of three diseases: Alzheimer’s disease, cerebrovascular disease, and Lewy body disease. But these three disease pathologies account for less than half of the variance in age-related cognitive decline (Boyle et al., 2013). More recently, hippocampal sclerosis, defined as severe gliosis and neuronal loss in the hippocampus and subiculum, has been increasingly studied in dementia and has been linked to Alzheimer’s disease (Attems and Jellinger, 2006); and is often studied in combination with TDP-43, as most cases with hippocampal sclerosis have TDP-43 pathology (Amador-Ortiz et al., 2007; Nelson et al., 2011; Zarow et al., 2012; Nag et al., 2015). Conversely, many cases of TDP-43 proteinopathy do not have hippocampal sclerosis. Here we perform a comprehensive study of TDP-43 pathology, Alzheimer’s disease pathology, and clinical Alzheimer’s-type dementia in a large autopsy cohort derived from a community cohort. We tested the hypotheses that TDP-43 mixed with Alzheimer's disease pathology is one of the most highly prevalent mixed pathologies; that it increases the odds of expressing clinical Alzheimer's-type dementia independent of pathological Alzheimer's disease, cerebrovascular disease (infarcts and small vessel disease–specifically arteriolosclerosis), Lewy body, and hippocampal sclerosis pathology; that it is more strongly related to clinical Alzheimer's-type dementia than the presence of pathological Alzheimer's disease alone; and that the TDP-43 pathological stage is an important determinant of clinical Alzheimer's-type dementia. Data came from two clinicopathological cohorts of ageing and dementia, the Memory and Aging Project and the Religious Orders Study, involving community-dwelling participants without dementia at study entry who agreed to annual cognitive testing and organ donation at death.

Materials and methods

Results

TDP-43 with and without pathologic Alzheimer's diagnosis, by clinical Alzheimer's dementia status

About 42% of the cohort had clinical Alzheimer’s-type dementia at the time of death (Table 1). The clinical diagnosis of Alzheimer’s-type dementia was pathologically confirmed as Alzheimer’s disease (NIA-Reagan intermediate or high) in over 80% of these 398 subjects. Interestingly, though Alzheimer’s disease was the most common pathologic diagnosis in subjects with clinical Alzheimer’s-type dementia, this was followed by TDP-43 pathology, which was present in the brains of almost two-thirds of those with clinical Alzheimer’s-type dementia.

In the 322 subjects with clinical Alzheimer’s-type dementia who had pathologically confirmed Alzheimer’s disease, mixed pathologies were much more common than single or ‘pure’ pathologies, and mixed Alzheimer’s disease and TDP-43 was the most common category of mixed pathologies. Figure 1 displays the proportion of mixed Alzheimer’s disease and TDP-43 pathologies in the brains of older subjects both with and without clinical Alzheimer’s-type dementia. Figure 1 shows that more than half of subjects with clinical Alzheimer’s-type dementia have mixed Alzheimer’s disease and TDP-43 pathologies, and that more than half of subjects in this group with Alzheimer’s disease pathologies also had TDP-43 pathology in their brain. In subjects with clinical Alzheimer’s-type dementia, a pathologic diagnosis of Alzheimer’s disease was more likely to be mixed with TDP-43 pathology (66.2%) than with infarcts (54.4%), arteriolosclerosis (34.8%), Lewy bodies (34.0%), or hippocampal sclerosis (19.9%). It was also much more common than ‘pure’ Alzheimer’s disease, defined as those with a pathologic diagnosis of Alzheimer’s disease without TDP-43, infarcts, arteriolosclerosis, Lewy bodies, or hippocampal sclerosis. Overall, the proportion of subjects with clinical Alzheimer’s-type dementia formerly labelled ‘pure pathologic Alzheimer’s disease’ was reduced by 60% from 16.1% to 6.5% when including TDP-43 as a mixed pathology. Conversely, the proportion of mixed pathologies in those with pathologically confirmed clinical Alzheimer’s-type dementia increased from 83.5% to 93.5% when considering TDP-43.

Open in a separate window

Figure 1

Mixed Alzheimer’s disease and TDP-43 pathologies by clinical Alzheimer’s-type dementia status proximate to death. Red = presence of pathologic diagnosis of Alzheimer’s disease (AD) without TDP-43. Blue = presence of pathologic transactive response DNA-binding protein 43 (TDP-43) without AD. Purple = presence of both AD and TDP-43. White = absence of AD and TDP-43.

TDP-43 was also present in those without clinical Alzheimer’s-type dementia (43.6%) and was generally at a lower stage; it extended out of the amygdala in just over half of those who had TDP-43 (54.0%), compared to 78.0% of those who had TDP-43 and had clinical Alzheimer’s-type dementia. In subjects without clinical Alzheimer’s-type dementia but with a pathologic diagnosis of Alzheimer’s disease, 49.4% had Alzheimer’s disease mixed with TDP-43 and again the TDP-43 tended to be at a lower stage. When separating those without clinical Alzheimer’s-type dementia into those with mild cognitive impairment versus those with no cognitive impairment, TDP-43 was slightly more prevalent in the brains of those with mild cognitive impairment (49.2% versus 39.1% in those with no impairment, P = 0.018).

TDP-43 with and without a pathologic diagnosis of Alzheimer’s disease in the clinical expression of Alzheimer’s-type dementia

We used a series of multiple logistic regression models adjusted for age, sex, and education to determine how TDP-43 pathology contributes to the clinical expression of clinical Alzheimer’s-type dementia, both independently and in combination with Alzheimer’s disease pathology. We first fit a model with a term for pathologic diagnosis of Alzheimer’s disease alone and as expected, a pathologic diagnosis of Alzheimer’s disease was associated with an increase in odds of clinical Alzheimer’s-type dementia (Table 3, Model 1). We then fit a model with TDP-43 pathology alone; TDP-43 was associated with more than double the odds of clinical Alzheimer’s-type dementia compared to no TDP-43 (Table 3, Model 2). In a model with both pathologic diagnosis of Alzheimer’s disease and TDP-43 pathology, TDP-43 pathology was associated with double the odds of clinical Alzheimer’s-type dementia independent of Alzheimer’s disease pathology (Table 3, Model 3). We then added terms for Lewy bodies, infarcts, arteriolosclerosis, and hippocampal sclerosis pathologies; the association of TDP-43 with clinical Alzheimer’s-type dementia was attenuated to a 51% increase in odds but still statistically significant independent of all other dementia-related pathologies (Table 3, Model 4). In a model adding an interaction term for TDP-43 and pathologic diagnosis of Alzheimer’s disease, the interaction was not significant (P = 0.69) suggesting there is an additive but not multiplicative effect of Alzheimer’s disease pathology and TDP-43 on the expression of clinical Alzheimer’s-type dementia. To illustrate the additive effect of Alzheimer’s disease with TDP-43 pathology, we fit a model with terms for (i) Alzheimer’s disease and TDP-43 mixed pathologies; (ii) Alzheimer’s disease without TDP-43 pathology; and (iii) TDP-43 without Alzheimer’s disease pathology, as well as terms for infarcts, arteriolosclerosis, Lewy bodies, and hippocampal sclerosis. As shown in Fig. 2, pathologic Alzheimer’s disease mixed with TDP-43 pathologies was associated with a much higher odds of clinical Alzheimer’s-type dementia [odds ratio (OR) = 6.73, 95% confidence interval (CI) = 4.18, 10.85] than pathologic Alzheimer’s disease without TDP-43 (OR = 4.62, 95% CI = 2.84, 7.52) or TDP-43 alone (OR = 1.67, 95% CI = 0.94, 2.94). We also fit an ordinal logistic regression model to test the association of these terms with any cognitive impairment (i.e. three-level ordinal outcome: clinical Alzheimer’s-type dementia, and mild cognitive impairment or clinical Alzheimer’s-type dementia compared to no cognitive impairment as the reference category) and the results were similar: pathologic Alzheimer’s disease mixed with TDP-43 pathologies was associated with a much higher odds of cognitive impairment (OR = 5.01, 95% CI = 3.45, 7.28) than pathological Alzheimer’s disease without TDP-43 (OR = 3.14, 95% CI = 2.16, 4.55) or TDP-43 alone (OR = 1.24, 95% CI = 0.81, 1.90).

Open in a separate window

Figure 2

Odds ratios for clinical Alzheimer’s-type dementia. Odds ratios provided for pathologic diagnosis of Alzheimer’s disease (AD) mixed with TDP-43 pathology, for pathologic diagnosis of AD without TDP-43 pathology, and for TDP-43 pathology without pathologic AD compared to not having pathologic diagnosis of AD or TDP-43 pathology, from a model adjusted for age, sex, education, cerebral infarcts, arteriolosclerosis, Lewy bodies, and hippocampal sclerosis.

Table 3

Odds ratios and 95% confidence intervals for clinical Alzheimer’s-type dementia (adjusted for age, sex, and education; n = 946)

Pathology type	Model 1	Model 2	Model 3	Model 4
TDP-43		2.07 (1.58, 2.72)	1.91 (1.44, 2.54)	1.51 (1.11, 2.05)
Pathologic diagnosis of Alzheimer’s disease	3.88 (2.86, 5.27)		3.71 (2.73, 5.06)	4.30 (3.08, 6.01)
Cerebral infarcts				1.69 (1.25, 2.29)
Arteriolosclerosis				1.51 (1.10, 2.09)
Lewy bodies				2.54 (1.81, 3.58)
Hippocampal sclerosis				4.37 (2.55, 7.48)

Open in a separate window

Values are OR (95% CI).

Finally, we examined the odds of clinical Alzheimer’s-type dementia by stage of TDP-43 pathology progression. In a model with terms for specific TDP-43 stage, as well as terms for Alzheimer’s disease, infarcts, arteriolosclerosis, Lewy bodies, and hippocampal sclerosis pathologies, a dose-response relationship was observed for odds of clinical Alzheimer’s-type dementia by increasing progression of TDP-43. As shown in Table 4, a significant association with odds of clinical Alzheimer’s-type dementia was only observed starting at stage 2 (extension beyond the amygdala to hippocampus and/or entorhinal cortex). Finally, in a model with the 6-point scale of TDP-43 severity as the term for TDP-43, each additional point of TDP-43 severity was associated with increased odds of clinical Alzheimer’s-type dementia (OR = 1.58, 95% CI = 1.31, 1.89).

Table 4

Odds ratio for clinical Alzheimer’s-type dementia comparing each stage of TDP-43 progression to no TDP-43, adjusted for age, sex, education, pathologic Alzheimer’s disease, infarcts, arteriolosclerosis, Lewy bodies, and hippocampal sclerosis pathologies

TDP-43 stage	OR (95% CI)
1: Amygdala	0.89 (0.58, 1.35)
2: Hippocampus/ entorhinal cortex	1.84 (1.25, 2.71)
3: Neocortex	2.44 (1.48, 4.02)

Open in a separate window

Discussion

In a cohort of almost 1000 older community-dwelling subjects we performed a comprehensive study of TDP-43 pathology, TDP-43 pathology mixed with a pathologic diagnosis of Alzheimer’s disease and the roles of these pathologies in the clinical expression of Alzheimer’s-type dementia. TDP-43 pathology was very common in the brains of older subjects, present in over half of all brains, and much more common in conjunction with other neurodegenerative pathologies, including not only Alzheimer’s disease but also hippocampal sclerosis (where it is extraordinarily common), arteriolosclerosis, and Lewy body disease. TDP-43 pathology was present in almost two-thirds of those with clinical Alzheimer’s-type dementia, being the second most common pathology after a pathologic diagnosis of Alzheimer’s disease. A pathologic diagnosis of Alzheimer’s disease mixed with TDP-43 was also the most common mixed pathology in subjects with a pathologic diagnosis of Alzheimer’s disease (compared to pathological Alzheimer’s disease mixed with infarcts, arteriolosclerosis, Lewy bodies or hippocampal sclerosis). We found no evidence of a multiplicative effect between TDP-43 pathology and a pathologic diagnosis of Alzheimer’s disease on the odds of clinical Alzheimer’s-type dementia; rather, TDP-43 acted as an additional additive pathology similar to other age-related pathologies. A pathologic diagnosis of Alzheimer’s disease mixed with TDP-43 pathology was associated with much higher odds of dementia than a pathologic diagnosis of Alzheimer’s disease without TDP-43, independent of other age-related pathologies. Finally, we found a dose-response relationship between TDP-43 pathologic stage and clinical Alzheimer’s-type dementia such that TDP-43 pathology needed to progress from the amygdala to the hippocampus/entorhinal cortex to have an association with clinical Alzheimer’s-type dementia.

TDP-43 pathology has a short but significant historical footprint in the field of Alzheimer’s disease and dementia (Wilson et al., 2011; Youmans and Wolozin, 2012; Josephs et al., 2014b; Josephs and Nelson, 2015; Chang et al., 2016). It was first recognized as a pathology that accumulated in the brains of patients with FTLD/ALS (Arai et al., 2006; Neumann et al., 2006), but has since been described as common in the brains of patients with Alzheimer’s disease (Amador-Ortiz et al., 2007; Hu et al., 2008; Josephs et al., 2014b), as well as hippocampal sclerosis (Amador-Ortiz et al., 2007; Nag et al., 2015; Cykowski et al., 2016), and possibly other neurodegenerative diseases (Higashi et al., 2007; Arai et al., 2009). For the most part, previous ageing and Alzheimer’s disease research has focused on TDP-43 pathology in those with clinical or pathological Alzheimer’s disease, with research participants identified from clinic samples or patient registries. The proportion of pathologically confirmed Alzheimer’s disease cases that had TDP-43 proteinopathy has been shown to range from between 19 and 57% in studies of 50 or more subjects (Amador-Ortiz et al., 2007; Uryu et al., 2008; Arai et al., 2009; Kadokura et al., 2009; Bigio et al., 2010; Davidson et al., 2011; Rauramaa et al., 2011; Josephs et al., 2014a), including 27% in a population-based sample (Keage et al., 2014). Our study, using community-based cohort of both subjects with and without clinical Alzheimer’s-type dementia, showed 58.5% of subjects with a pathologic diagnosis of Alzheimer’s disease had TDP-43 proteinopathy. In studies involving only subjects clinically diagnosed with clinical Alzheimer’s-type dementia or any dementia, the proportion with TDP-43 was lower (29 to 41%) (Davidson et al., 2011; Robinson et al., 2011) than the proportion that we observed. Our study may have a larger proportion of TDP-43 positivity than some other studies because we investigated the amygdala which has been noted as the most commonly involved region in some studies including the current one (Higashi et al., 2007; Hu et al., 2008).

Previous work has shown that TDP-43 pathology is important in dementia, lowers cognition and accelerates cognitive decline. A study using a subset of an autopsy cohort found an association between TDP-43 and cognitive impairment, independent of Alzheimer’s disease pathology or hippocampal sclerosis (Nelson et al., 2010). In the Religious Orders Study, investigators previously demonstrated that TDP-43 positivity is higher in clinical Alzheimer’s-type dementia than in normal cognition, with mild cognitive impairment at an intermediate level (Tremblay et al., 2011) and that TDP-43 was related to more rapid cognitive decline and a higher likelihood of dementia (but not mild cognitive impairment, though the sample size was much smaller than the current study) and accounted for nearly as much variability in cognitive decline as did tau tangles (Wilson et al., 2013). Another prospective population-based study in the oldest-old also found an association with dementia (Robinson et al., 2011). Studies in pathologically confirmed Alzheimer’s dementia have shown TDP-43 related to lower cognition (Josephs et al., 2014b). Importantly, the current study extends these findings to a large, community-based cohort with and without clinical Alzheimer’s-type dementia, and shows that a pathologic diagnosis of Alzheimer’s disease mixed with TDP-43 pathology was associated with much higher odds for clinical Alzheimer’s-type dementia than Alzheimer’s disease pathology without TDP-43, after accounting for other age-related pathologies. It has been long recognized that subjects with a pathologic diagnosis of Alzheimer’s disease may not exhibit dementia. The current study extends the data on mixed Alzheimer’s disease pathologies, and indicates that TDP-43 is another important mixed pathology linked to crossing the threshold to dementia in subjects with a pathologic diagnosis of Alzheimer’s disease (Schneider et al., 2007). TDP-43 may therefore help to explain some of the unexplained variance accounted for by age-related pathologies in the expression of dementia and cognitive impairment in later life (Boyle et al., 2013).

There is increasing recognition of the role of mixed or comorbid pathologies in the expression of clinical Alzheimer’s-type dementia (Schneider et al., 2007). Almost all of the subjects with clinical Alzheimer’s-type dementia in this cohort had two or more pathologies when considering TDP-43 in addition to the other most commonly studied age-related pathologies. Overall, <10% of subjects with clinical Alzheimer’s-type dementia were left with ‘pure Alzheimer’s disease’ when TDP-43 was taken into account, highlighting that the clinical syndrome often assumed to be the sole manifestation of Alzheimer’s disease pathology is influenced by numerous other age-related pathologies (Jellinger and Attems, 2015). This has public health implications for treatment and research on dementia, as innovations on interventions for alternate pathologies such as TDP-43 could help to alleviate the societal burden of Alzheimer’s dementia even in the absence of prevention or treatment of the amyloid plaques and PHF-tau tangles that are the hallmarks of Alzheimer’s disease pathology (Chang et al., 2016). Moreover, the dose-response relationship between TDP-43 stage progression and Alzheimer’s disease-dementia, along with the findings that TDP-43 is present in lower stages in the brains of people with no Alzheimer’s disease pathology and in those without clinical Alzheimer’s-type dementia, suggest there may be a critical window for intervention prior to the point where the progression of Alzheimer’s disease and TDP-43 pathology manifest as dementia. From a research perspective, it remains unclear why some subjects with amyloid-β and tau tangles exhibit TDP-43 while others do not. Genetic factors, e.g. variants in TMEM106B, are likely to be factors (Yu et al., 2015). Whether genetic interactions or pathologic triggers (e.g. amyloid-β deposition) also play a role is not clear (Youmans and Wolozin, 2012; Chang et al., 2016). Clearly, mixed neurodegenerative pathologies and the relationship between specific comorbid pathologies and dementia deserve further study.

This study extends previous research seeking to develop a staging scheme for TDP-43 in Alzheimer’s disease. Josephs et al. (2014a) originally developed a scheme identifying five stages of progression of TDP-43 deposition in Alzheimer’s disease and demonstrated a relationship between higher TDP-43 stage and lower cognitive ability. Recently, the investigators extended this staging scheme to include limbic regions of the brain, identifying six stages of TDP-43 deposition, progressing from the amygdala (stage 1), to entorhinal cortex and subiculum (stage 2), to the dentate gyrus of the hippocampus and occipitotemporal cortex (stage 3) to insular cortex, ventral striatum, basal forebrain and inferior temporal cortex (stage 4), to substantia nigra, inferior olive and midbrain tectum (stage 5), and to basal ganglia and middle frontal cortex (stage 6) (Josephs et al., 2016). Our staging scheme recognized only four stages (including no TDP-43), but the progression from amygdala to hippocampus and entorhinal cortex to neocortex followed the same order as the Josephs et al. (2014a) staging. Uniquely, our staging scheme was rendered in participants without regard to pathological or clinical Alzheimer’s disease status. Our study thus extends previous research by showing a relationship between TDP-43 stage and pathological likelihood of Alzheimer’s disease, as well as between TDP-43 stage and likelihood of clinical Alzheimer’s-type dementia. A novel finding was that TDP-43 had to progress to stage 2 (hippocampus and entorhinal cortex) before a relationship with clinical Alzheimer’s-type dementia was observed.

Strengths of this study include the use of autopsy data from a large, community-based cohort of older subjects, some of whom developed clinical Alzheimer’s-type dementia prior to death over the course of follow-up and some of whom did not. Findings are therefore more generalizable to the general population of older adults than a memory clinic-based sample, or a brain bank of brains with confirmed Alzheimer’s disease. Further, annual follow-up and autopsy rates were high, minimizing the risk for bias due to selective attrition before autopsy, and the time between last clinical evaluation and autopsy was relatively short. Uniform clinical evaluation and established criteria were used to clinically classify participants. Neuropathological evaluations were conducted blinded to clinical evaluations and diagnoses. Limitations include a relatively healthy and racially homogeneous volunteer cohort that agreed to autopsy, potentially limiting generalizability. For this analysis, pathologies like TDP-43 without clearly defined pathologic diagnoses were dichotomized to facilitate comparisons with pathologic diagnosis of Alzheimer’s disease and across pathologies; this dichotomizaton was data-driven, but using different thresholds for presence of pathology could affect our results. Autopsy procedures were carried out on only one hemisphere of the brain, raising the possibility that certain pathologies that present unilaterally were not observed. We did not include data on TDP-43 neurites, which may lead us to underestimate the per cent of mixed Alzheimer’s disease; further study of TDP-43 neurites in relation to Alzheimer’s disease and other pathologies is needed. At the time of analyses, the caudate had not been assessed for TDP-43 in the full cohort, limiting our ability to compare to other TDP-43 staging schemes that involve the caudate (Josephs et al., 2014a, 2016). More research is needed to understand how TDP-43 contributes to dementia in the context of Alzheimer’s disease and other known neuropathologies.

Acknowledgements

Glossary

References