TDP-43 Is Associated with Subiculum and Cornu Ammonis 1 Hippocampal Subfield Atrophy in Primary Age-Related Tauopathy

INTRODUCTION

Primary age-related tauopathy (PART) describes a pathology that combines a limbic predominant distribution of neurofibrillary tangles (NFTs) in the absence or limited distribution of amyloid-β (Aβ) [1]. PART is diagnosed when the Braak NFT stage [2] is I-V and the Aβ Thal phase is 0 (Definite PART) or Aβ Thal phase is 1–2 (Possible PART) [3]. PART has been associated with variable cognition ranging from normal cognition to dementia, with the latter previously being referred to as ‘tangle-predominant senile dementia’ (TPSD), ‘tangle-only dementia’, ‘preferential development of NFT without senile plaques’, and ‘senile dementia of the neurofibrillary tangle type’ (SD-NFT) [1]. Studies show that patients with PART have atrophy of the hippocampus which was presumed to be due to accumulation of the protein tau in the form of NFTs [4–6].

The transactive response DNA-binding protein of 43 kDa (TDP-43) is present in different neurodegenerative diseases. It was first linked to frontotemporal lobar degeneration and amyotrophic lateral sclerosis [7], but later was shown to be present in a significant percentage of cases of with Alzheimer’s disease, as well as in other neurodegenerative diseases [8–10]; some researchers have advocated for the label limbic predominant age-related TDP-43 encephalopathy neuropathologic changes or LATE-NC [11]. In recent studies, we reported finding an association between TDP-43 and whole hippocampal volume loss in cases of PART [12, 13].

Studies on Alzheimer’s disease have shown that TDP-43 is associated with volume loss of specific hippocampal subfields, which together contributes to whole hippocampal volume loss [2, 15]. In addition, TDP-43 burden has been shown to be associated with inward deformation of the hippocampus in zones approximating CA1 and subiculum in AD [16]. In PART, similar to Alzheimer’s disease [9, 17], TDP-43 pathology starts in the medial temporal lobe and progresses to the hippocampus and neocortex [2]. Although there are studies that have investigated the distribution and burden of TDP-43 pathology in hippocampal subfields, volume loss in specific subfields of the hippocampus associated with TDP-43 in PART have not been investigated.

Consequently, the lack of knowledge of this relationship prompted us to investigate the associations between hippocampal subfields and TDP-43 in PART, to ascertain the specific hippocampal subfields that contribute to the association between TDP-43 and the overall loss of hippocampal volume. Results from this study will increase our understanding of TDP-43 and PART pathophysiology and highlight specific hippocampal subfields as potential targeted imaging biomarkers of TDP-43 pathology. We hypothesized that patients with PART and TDP-43 will exhibit smaller volumes in the CA1, subiculum and dentate gyrus of the hippocampus since these are the areas primarily affected by TDP-43 deposition.

METHODS

Brain MRI segmentation

In this study, we used the MRI scan nearest to the time of death in participants with multiple serial MRIs. All participants underwent a volumetric MRI using a standardized protocol [21], and all scans underwent correction for intensity inhomogeneity and were subjected to quality control for subsequent analysis. The segmentation of hippocampal subfields was performed using FreeSurfer version 7.4.1, which utilizes a probabilistic atlas constructed from ultra-high-resolution ex vivo MRI data (approximately 0.1 mm isotropic) to produce an automated segmentation of the hippocampal substructures [22, 23]. FreeSurfer subdivides hippocampal substructures into head, body, and tail, where applicable, and generates an additional set of segmentations of each hemisphere at different levels of hierarchy. This approach echoes historical methodologies, as delineated by an early foundational paper by Clifford et al., which pioneered the analysis of the hippocampus by its segmented region [24]. An example segmentation is shown in Figs. 1and 2. Averaging the volumetric data of the left and right hippocampi has been utilized to reduce lateralization bias, to increase statistical power by reducing variability due to side-specific differences, and for a holistic representation of hippocampal volume changes. To adjust for differences in head size, total intracranial volume (TIV) was calculated using the standard FreeSurfer processing pipeline.

OPEN IN VIEWER

Fig. 1

Hippocampal subfields’ sagittal view. GC-ML-DG, Granule Cell and Molecular Layers of the Dentate Gyrus; HATA, Hippocampal-Amygdala Transition Area.

OPEN IN VIEWER

Fig. 2

Hippocampal subfields’ coronal view.

RESULTS

Demographic characteristics are displayed in Table 1. Of the 115 participants, approximately 43% were female. The average age at the time of their last MRI was 84 years, and the average age at death was 88 years. Thirty-seven participants were TDP-positive (32%). TDP-43 type was available for 15 participants in the TDP-43 positive group, and all were type-alpha. There was a significant difference observed between TDP-positive and negative groups in the frequency of hippocampal sclerosis (p = 0.001). Ten participants had hippocampal sclerosis (8.7%), with eight in the TDP-positive group (21.62%) and two in the TDP-negative group (2.56%). Both patients in the TDP-43 negative group had vascular lesions in the hippocampus. No significant differences were noted in the remaining demographic variables between TDP-positive and TDP-negative groups. There was a trend, however, for the TDP-43-positive group to be older at the time of death.

The findings for hippocampal volumes are shown in Table 2. Whole hippocampal volume was significantly smaller in the TDP-positive cases compared to the TDP-negative cases (p = 0.049). After stratification of the hippocampus into head, body, and tail, the TDP-positive cases showed smaller volumes of the hippocampal body (p = 0.031) and tail (p = 0.021), but not the hippocampal head (p = 0.129), compared to TDP-negative cases. When these regions were further subdivided into hippocampal subfields, the TDP-positive cases showed smaller volumes of molecular layer of hippocampal body (p = 0.046), CA1 body (p = 0.031), subiculum body (p = 0.023), as well as presubiculum body (p = 0.046) compared to TDP-negative cases. After separating definite from possible PART, no association was found between TDP-43 and volume in those with definite PART. TDP-43 staging was available for 32 participants of the TDP-43 positive group and there was no correlation found between staging and volumes of the whole hippocampi or subfields.

Results from the logistic regression analysis, are presented in Table 3. After adjusting for TIV, we found that reduced volumes of the whole hippocampus, hippocampal body, and its constituents including the molecular layer, CA1, subiculum and presubiculum had higher odds of being TDP-43 positive. After adjusting for hippocampal sclerosis, the results were essentially unchanged. TDP-43 remained associated with hippocampal body (p = 0.046), hippocampal tail (p = 0.025), CA1 body (p = 0.041), and subiculum body (p = 0.036), and approached significance with whole hippocampus (p = 0.055).

DISCUSSION

In this cohort of 115 cases of PART, we assessed the association of TDP-43 status with hippocampal subfield atrophy. We found that TDP-positive cases exhibited smaller whole hippocampal volumes compared to TDP-negative cases, validating our previous results [12, 13]. Further stratification revealed that the body and tail of the hippocampus showed smaller volumes in those with TDP-43. By further dissecting the hippocampus into subfields, we were able to attribute volume loss to the molecular layer of hippocampal body, CA1 body, subiculum body, and presubiculum body.

Our finding of smaller whole hippocampal volume in TDP-positive cases with PART concurs with our previous studies demonstrating the influence of TDP-43 on brain volumes in PART [12, 13]. In this study, we further investigate the specific regions and subfields of the hippocampus and discovered that differences in volume loss are more prominent in the body and tail of the hippocampus in the TDP-positive cases. Moreover, with further stratification and analysis, volume loss was observed in the molecular layer of hippocampal body, CA1 body, subiculum body, as well as presubiculum body. To our knowledge, this is the first study to investigate the effects of TDP-43 in patients with PART on hippocampal subfields. Based on the FreeSurfer anatomical atlas [22], we note that the hippocampal tail has no subfield components. However, the body is comprised of various components, including the molecular layer, CA1, subiculum, and presubiculum, which are all significantly smaller in volume in TDP-positive cases. Notably, the same effect was not observed in absence of beta-amyloid which is consistent with our previous study that also found no association between TDP-43 and whole hippocampal volumes in patients with definite PART [5]. This is likely because the vast majority of patients with definite PART are cognitively normal with absent to minimal volume loss, and no range of data for volumes. We did observe a trend for the volume of the head of the hippocampus to be smaller in the TDP-positive cases compared to the TDP-43 negative cases.

In neuropathological studies where TDP-43 pathology has been observed, there is significant deposition of TDP-43 in regions and subregions of medial temporal lobe structures including the CA1 and subiculum regions of the hippocampus [25]. TDP-43 has been previously found to have an effect on the hippocampal subfields CA1 and subiculum in Alzheimer’s disease and has been associated with inward deformation of the hippocampus, which correlates with cognition scores [14]. It is not surprising that these two regions show volume loss associated with TDP-43 in Alzheimer’s disease and now in PART, because these two regions are directly affected by TDP-43 deposition [17]. In PART, however, it is vulnerability of the CA2 subregion due to NFT degeneration that is greatest [4, 26]. Our findings therefore are more keeping with volume loss due to TDP-43 than volume loss due to NFTs and tau.

We discovered that TDP-43 in PART predominantly impacts the hippocampal body, specifically affecting four main areas: the molecular layer, CA1, subiculum, and presubiculum. Given the observed volume loss in these areas, it may be worth exploring their potential as imaging biomarkers for TDP-43 associated atrophy in future longitudinal studies. However, the long-term performance and reliability of these biomarkers remain to be established through comprehensive research. Differential atrophy of the CA1 versus CA2 regions of the hippocampus could be utilized to assess treatment response to a tau versus TDP-43 targeted pharmacological molecule.

The CA1 and subiculum are also involved in hippocampal atrophy in Alzheimer’s disease where both Aβ and tau are present [27–29]. In individuals with Alzheimer’s disease, reduced volumes of both CA1 and subiculum have been observed, consistent with atrophy in these two subfields [30]. Given that we have demonstrated a similar pattern of involvement in PART that is related to TDP-43, one could hypothesize that it is TDP-43 that is possibly driving atrophy in these two subfields in Alzheimer’s disease. It is unclear however, whether there are differences between TDP-43 related subfield volume loss in Alzheimer’s disease and TDP-43 related subfield volume loss in PART. For example, our analysis suggests more specific involvement of these regions in the body of the hippocampus as opposed to the head. It is possible that the hippocampal head is more affected than the body in Alzheimer’s disease.

Unlike the CA1 and subiculum which have been discussed in previous studies, less is known about the molecular layer. In one study comparing the volume of the molecular layer in cognitively intact versus mild cognitively impaired patients, no significant difference was observed [31]. Similarly, little has been reported on the presubiculum in the body of the hippocampus.

There is a strong association between TDP-43 pathology and hippocampal sclerosis, with hippocampal sclerosis conceptualized as a downstream consequence of the presence of TDP-43 [32]. Hence, it is not surprising that those with TDP-43 were more likely to have hippocampal sclerosis compared to those without TDP-43. Unlike hippocampal sclerosis, there was no difference in the Braak NFT stage or the Thal phase between those with and without TDP-43. This is most likely due to the fact that our inclusion criteria limited cases to specific Braak and Thal stages. It is well known that patients with TDP-43 have higher Braak NFT stages when the full spectrum of Braak stage cases (I–VI) are included in the study [33, 34].

AUTHOR CONTRIBUTIONS

Hossam Youssef (Conceptualization; Data curation; Formal analysis; Methodology; Project administration; Visualization; Writing – original draft); Rodolfo G. Gatto (Validation; Writing – review & editing); Nha Trang Thu Pham (Methodology; Resources; Software); Ronald C. Petersen (Validation; Writing – review & editing); Mary M. Machulda (Validation; Writing – review & editing); R. Ross Reichard (Validation); Dennis W. Dickson (Validation); Clifford R Jack (Validation; Writing – review & editing); Jennifer L. Whitwell (Conceptualization; Supervision; Validation; Writing – review & editing); Keith A. Josephs (Conceptualization; Methodology; Supervision; Validation; Writing – review & editing).