TDP-43 pathology effect on volume and flortaucipir uptake in Alzheimer’s disease

Abstract

INTRODUCTION:

Alzheimer’s disease (AD) patients $\ge$70 years show smaller medial temporal volumes despite less ¹⁸F-flortaucipir-positron emission tomography (PET) uptake than younger counterparts. We investigated whether TAR DNA-binding protein 43(TDP-43) was contributing to this volume-uptake mismatch.

METHODS:

Seventy-seven participants with flortaucipir-PET and volumetric magnetic resonance imaging underwent postmortem AD and TDP-43 pathology assessments. Bivariate-response linear regression estimated the effect of age and TDP-43 pathology on volume and/or flortaucipir standardized uptake volume ratios of the hippocampus, amygdala, entorhinal, inferior temporal, and midfrontal cortices.

RESULTS:

Older participants had lower hippocampal volumes and overall flortaucipir uptake. TDP-43-immunoreactivity correlated with reduced medial temporal volumes but was unrelated to flortaucipir uptake. TDP-43 effect size was consistent across the age spectrum. However, at older ages, the cohort mean volumes moved towards those of TDP-43-positives, reflecting the increasing TDP-43 pathology frequency with age.

DISCUSSION:

TDP-43 pathology is a relevant contributor driving the volume-uptake mismatch in older AD participants.

1. BACKGROUND

Alzheimer’s Disease (AD) is a neurodegenerative disease characterized by the deposition of $\mathrm{\beta }$-amyloid and tau-immunoreactive plaques and neurofibrillary tangles (NFT).^1,2 Imaging biomarkers, such as structural magnetic resonance imaging (MRI) to measure atrophy and positron emission tomography (PET) radiotracers for $\mathrm{\beta }$-amyloid and tau burden, are being utilized more often to uncover the pathogenesis of AD. While an amyloid-centric pathogenesis was initially widely accepted,³ neuroimaging and pathologic studies have shown that tau pathology also plays a key role in the development of atrophy and clinical presentation in AD.^4–7 The distinction between older-onset and younger-onset phenotypes, for example, appears to be governed by the pattern of regional atrophy and NFT burden, with older AD patients having more pronounced medial temporal lobe involvement and younger onset patients showing more prominent cortical degeneration.^8–11 Nonetheless, AD pathology rarely occurs unaccompanied by other pathologies, including Lewy body disease, vascular pathology, and TAR DNA-binding protein 43 (TDP-43) proteinopathy.^12–14 The contribution of these co-pathologies to the neurodegeneration and clinical presentations in AD is a current knowledge gap and research focus.

The tau radioligand ¹⁸F-flortaucipir has high specificity for the mixed 3-/4-repeat tau found in AD.¹⁵ In a previous study,⁸ we had found that while all younger AD participants showed high neocortical but low entorhinal cortical flortaucipir retention regardless of age, the older AD participants displayed two patterns of flortaucipir uptake that were age-dependent: low uptake in both entorhinal cortex and neocortex in older participants (median age at scan:76 years) and high uptake in both regions in younger participants (median age at scan:62). In a successive study,¹⁶ we assessed how these differences were related to volume. We found that younger typical AD participants (aged 50–69) showed strong correlations between flortaucipir uptake and volume within the hippocampus and entorhinal cortex. However, we found that the older AD participants (aged$\ge$70), while having the greatest volume loss within the hippocampus and entorhinal cortex, surprisingly had relatively lower flortaucipir uptake compared to younger counterparts. We hypothesized that the greater medial temporal volume loss despite lower flortaucipir retention in these older participants could be related to another pathology –TDP-43 pathology– having an additive effect. TDP-43 is a nuclear protein whose nuclear loss and cytoplasmic aggregation in neurons and glia occurs in many neurodegenerative diseases, including AD.¹⁷ Moreover, TDP-43 proteinopathy appears more frequent in old age¹⁸ and associates with hippocampal volume loss.¹⁹

In the present study, we aimed to determine whether TDP-43 pathology could be contributing to the volume-uptake mismatch in older AD participants. We hypothesized that the presence of TDP-43 pathology would be associated with greater medial temporal lobe volume loss despite lower flortaucipir uptake, and that this association would be more pronounced with increasing age.

2. METHODS

1.1. Study Participants

All participants were prospectively recruited and followed in one of three NIH-funded enterprises: Mayo Clinic Study of Aging, Alzheimer’s Disease Research Center, or Neurodegenerative Research Group. Participants who had completed ¹⁸F-flortaucipir-PET and structural 3T MRI brain scans taken within a 48-hour period and had undergone a postmortem assessment of TDP-43 status and Braak NFT stage were included. Participants with a neuropathologic diagnosis of frontotemporal lobar degeneration with TDP-43 (FTLD-TDP) were excluded.²⁰ Seventy-seven participants were identified and included in the study. All had varying degrees of Alzheimer’s disease neuropathologic change (ADNC).²

This study was approved by the Mayo Clinic Institutional Review Board and all participants/proxies signed written consent.

2.2. Cognitive state evaluation

All participants underwent risk factor evaluations (including ApoE genotyping) and standardized neurological and neuropsychological battery, as well as additional laboratory and imaging studies, as previously reported.^21–23 A diagnosis of cognitive state (cognitively unimpaired, mild cognitive impairment, or dementia) was determined following a consensus meeting and was based on all available data.^21–24

2.3. Image acquisition, processing, and analyses

The 3T volumetric MRI protocol included a magnetization prepared rapid gradient echo (MPRAGE) sequence, while ¹⁸F-flortaucipir-PET scans were acquired using a PET/CT scanner (GE Healthcare, Milwaukee,Wisconsin) operating in a 3D mode, as previously described.¹⁶ Gray matter volumes and flortaucipir uptakes were assessed in five regions of interests (ROI) affected by TDP-43 pathology in AD¹⁴: amygdala, entorhinal cortex (ERC), hippocampus, interior temporal cortex and middle frontal cortex. The Mayo Clinic Adult Lifespan Template (MCALT) atlas²⁵ was used to calculate regional gray matter volumes. Total intracranial volume (TIV) was also calculated to correct for head size. For gray matter volumes, left and right volumes were summed and volumes adjusted for total TIV were calculated and converted to Z-scores reflecting deviation from the control average. The control group included 168 cognitively unimpaired participants aged 30–50 years who had been recruited into the Mayo Clinic Study of Aging²² and underwent the same imaging modalities and measurements as the main study participants. Since hippocampal flortaucipir measurements typically show low uptake and can also be confounded by off-target binding to the choroid plexus,^15,26 we decided to include other regions within the medial temporal lobe, such as the amygdala and the ERC. Median flortaucipir uptake was calculated across gray and white matter in each of the five ROIs, and standardized uptake value ratios (SUVRs) were successively generated as a weighted average of left and right median uptake values in each ROI divided by the uptake in the cerebellar crus gray matter, as previously described.²⁷ The principal analyses were performed without partial volume correction (PVC) of flortaucipir to maintain relative independence of flortaucipir and gray matter measurements. However, analysis of data corrected for partial volume effect using a two-compartment model was additionally performed to provide transparency.

2.4. Neuropathologic evaluation

All participants underwent postmortem examination of the brain. To assess the degree of AD pathology, the recommendations of the Consortium to Establish a Registry for Alzheimer’s disease (CERAD)²⁸ guidelines were followed. Each case was assigned a Braak and Braak NFT stage²⁹ using modified Bielschowsky silver stain, and a CERAD neuritic plaque stage.²⁸ A Thal phase for $\mathrm{\beta }$-amyloid plaques was also assigned.³⁰ All cases included satisfied the criteria for low, intermediate, or high probability AD² or for primary age related tauopathy (PART).^31,32

Since the amygdala is the earliest region affected by TDP-43 pathology in AD,^33,34 5-$\mathrm{\mu }$m-thick formalin-fixed paraffin-embedded amygdala sections immunostained for phosphorylated TDP-43 (pS409/410, mouse monoclonal, 1:5,000, Cosmo Bio, Tokyo,Japan) were screened. All cases showing either the typical TDP-43 inclusions as seen in FTLD³⁵ (type-α) and/or TDP-43 inclusions colocalizing with NFTs³⁶ (type-$\mathrm{\beta }$) within the parenchyma were considered TDP-43-positive(+).

Since vascular pathology is linked to medial temporal lobe and hippocampal atrophy,³⁷ our analyses had to be adjusted for it. Arteriolosclerosis and cerebral amyloid angiopathy (CAA) were graded from 0–3, for absent, mild, moderate, or severe. Microinfarcts and lacunar (<1cm)/large infarcts (>1cm) were graded dichotomously as 0=absent or 1=present. A composite vascular score was then determined as previously described.³⁸

2.5. Statistical analyses

All statistical analyses were performed using R version 3.6. Significance level was set at alpha=0.05. For each of the 5 ROIs, we fit a multivariate linear regression model with two response variables (volume adjusted for TIV and log-transformed flortaucipir SUVR). The predictors in each model included age at scan (centered at 75 years and divided by 10) and TDP-43 status, with composite vascular score and Braak NFT stage included as covariates. An interaction term between TDP-43 status and age at scan was included to assess whether the TDP-43 effect varied by age at scan. Since these interactions were not significant at p<0.05 for all ROIs, the interaction term was removed, and the model was rerun. Additionally, because the composite vascular score did not emerge as a significant individual predictor of either volume or flortaucipir uptake (p>0.05 in all ROIs), it was also dropped from the model. Therefore, the final models included TDP-43 status, age at scan, and Braak NFT stage.

We then summarized these multivariate models using hypothesis-error (HE) plots.³⁹ These HE plots show an error ellipse representing within group variations and superimposed lines representing the effects of TDP-43 status, age, and Braak NFT stage. The orientation of each line depicts the variable’s relationship to volume and tau uptake³⁹:a line that is generally horizontal indicates the variable has an effect on mean volume rather than tau uptake; a line that is generally vertical indicates the variable has an effect on mean tau uptake rather than volume; and a line that is generally diagonal indicates a variable has an effect on both mean volume and mean tau uptake. If the line does not extend beyond the error ellipse, the variable is not significant at p<0.05. We also performed a sensitivity analysis excluding 20 participants who had received a final clinical diagnosis of cognitively unimpaired or mild cognitive impairment to assess for any influence of cognitive state on our primary analyses. In a subsequent step, because the frequency of TDP-43 pathology increases with older age, we sought to understand how TDP-43 status influenced overall patterns in the cohort by comparing mean volume and tau uptake as a function of age by TDP-43 status in the cohort, as a whole, ignoring TDP-43 status (n=77) and in the subset who died with dementia ignoring TDP-43 status (n=57).

3. RESULTS

3.1. Participant characteristics

Of the 77 participants, 31 (40%) showed TDP-43-immunoreactivity. Demographic, clinical and neuropathologic features of the cohort are shown in Table 1. TDP-43(+) participants were older at the time of last MRI scan (82 versus 74 years) and of death (84 versus 76 years). Clinically, they were more likely to had been diagnosed with cognitive impairment (97% versus 76%). After adjusting for age at death, higher Braak NFT stages (V-VI), Thal β-amyloid phases (4–5), and composite vascular scores (score=4) were more frequently seen in those with TDP-43 pathology.

Table 1.

Demographic, clinical, and neuropathologic characteristics of participants stratified by TDP-43 status.

Characteristic	All (n = 77)	TDP− (n = 46 (60%))	TDP+ (n = 31 (40%))	p-value
Female, n (%)	30 (39%)	17 (37%)	13 (42%)	0.81
APOE ε4 carrier, n (%)	40 (53%)	25 (54%)	15 (52%)	>0.99
Education, yr	14 (8, 20)	14 (8, 20)	16 (8, 20)	0.78
Age at last scan, yr	76 (51, 98)	74 (51, 98)	82 (55, 92)	0.08
Age at death, yr	79 (51, 99)	76 (51, 99)	84 (60, 95)	0.07
Last tau to death, yr	2.4 (0.0, 5.1)	2.2 (0.0, 4.5)	2.7 (0.6, 5.1)	0.31
Last clinical diagnosis, n (%)				0.008
CU	12 (16%)	11 (24%)	1 (3%)
MCI	8 (10%)	2 (4%)	6 (19%)
Dementia	57 (74%)	33 (72%)	24 (77%)
Braak NFT stage, n (%)				0.01
0	3 (4%)	3 (7%)	0 (0%)
I	2 (3%)	1 (2%)	1 (3%)
II	7 (9%)	6 (13%)	1 (3%)
III	6 (8%)	6 (13%)	0 (0%)
IV	8 (10%)	6 (13%)	2 (6%)
V	21 (27%)	8 (17%)	13 (42%)
VI	30 (39%)	16 (35%)	14 (45%)
Thal amyloid phase, n (%)				0.004
0	6 (8%)	6 (13%)	0 (0%)
1	4 (5%)	3 (7%)	1 (3%)
2	5 (6%)	5 (11%)	0 (0%)
3	7 (9%)	5 (11%)	2 (6%)
4	17 (22%)	9 (20%)	8 (26%)
5	37 (48%)	18 (39%)	19 (61%)
Composite Vascular, n (%)				0.008
0	4 (5%)	3 (7%)	1 (3%)
1	7 (9%)	7 (16%)	0 (0%)
2	37 (50%)	23 (52%)	14 (47%)
3	5 (7%)	5 (11%)	0 (0%)
4	21 (28%)	6 (14%)	15 (50%)

Data shown are n (%) or median (range). For categorical variables, p-values are from Fisher’s Exact Test. For continuous variables, p-values are from Wilcoxon Rank Sum test. p-values for Braak NFT stage, Thal amyloid phase, and composite vascular are from logistic regression adjusting for age at death. p-values <0.05 are considered significant and are shown in bold italics.

Abbreviations: APOE = apolipoprotein E; CU = cognitively unimpaired; MCI = mild cognitive impairment; NFT = neurofibrillary tangle

3.2. Differences within age spectrum and TDP-43 status by region

The scatter plots in Figure 1 show the distribution of the data analyzed. The HE plots in Figure 2 show a general indication of how age at scan and TDP-43 status affected volume or flortaucipir uptake in the 5 ROIs. Age was associated with medial temporal lobe volume and flortaucipir uptake in both medial temporal lobe and neocortical areas after accounting for Braak NFT stage. Contrarily, TDP-43 status had effects limited to the volume dimension within the medial temporal lobe and had seemingly no effect on flortaucipir uptake. Results from analyses using data with partial volume correction were very similar (not shown).

Figure 1. Distribution of raw data stratified by TDP-43 status.

Relationships between age at scan and volume (left column), age and flortaucipir SUVRs (center column), and volume and flortaucipir uptake (right column) are shown. Spearman correlation coefficients indicating the strength and direction of correlations are also shown for all 5 ROIs. TDP-43(+) are represented by pink circles and TDP-43(−) by green circles. Abbreviations: frontal mid = middle frontal cortex; temporal inf = inferior temporal cortex; ROI = region of interest; SUVR = standardized uptake value ratio; vol adj TIV = volume adjusted for total intracranial volume

Figure 2. Overall effects of age at scan, TDP-43, and Braak stage on volume and flortaucipir uptake.

Results from analyses using data without partial volume correction are shown. Results from primary analyses using all participants (left) and from the sensitivity analyses using participants with dementia (right) are also shown. Significance scaling hypothesis error (HE) plots show an error ellipse, representing within group variations, and superimposed lines, representing the individual predictors (TDP-43 status in pink, age at scan in green, and Braak stage in yellow). If the lines extend outside the ellipse, the multivariate test is significant at p< 0.05. The direction of the lines depicts how the groups differ, i.e., if horizontal, group differences are in terms of volume; if vertical, differences are in terms of flortaucipir uptake; and if diagonal, difference are in terms of both. Relationships between volume and tau by age at scan and TDP-43 status are also shown, where pink triangles represent the TDP-43(+) and gray circles represent the TDP-43(−). Analyses using data with PVC yielded very similar results (not shown). Abbreviations: ERC = entorhinal cortex; Est vol adi TIV = estimated volume adjusted total intracranial volume; frontal mid = middle frontal cortex; PVC = partial volume correction; SUVR = standardized uptake value ratio; temporal inf = inferior temporal cortex

3.3. Age effect on volume and flortaucipir uptake

The effect sizes of age at scan on volume and flortaucipir uptake are shown in Figure 3A and Table 2.

Figure 3. Effects of age and TDP-43 status on volume and flortaucipir uptake.

Forest plots show the percentage of decreases or increases in volume or flortaucipir uptake as a function of age at scan (A) and TDP-43 status (B). Results from the primary analyses are represented by the thick black lines, those from the sensitivity analyses are represented by the gray dashed lines. The dots represent the estimates, the horizontal lines the confidence intervals. Confidence intervals not crossing the line of null effect (red dashed vertical line) are considered significant at p<0.05. Abbreviations: Frontal mid = middle frontal cortex; PVC = partial volume correction; SUVR = standardized uptake value ratio; Temporal inf = inferior temporal cortex

Table 2.

Estimates of predicted TDP−43 and age effects on volume and flortaucipir uptake.

Region	Age			TDP
	Volume	Tau (No PVC)	Tau (PVC)	Volume	Tau (No PVC)	Tau (PVC)
Amygdala	−0.5% (−4%, 3%)	−5% (−8%, −2%)	−6% (−9%, −2%)	−3% (−11%, 5%)	4% (−3%, 12%)	6% (−2%, 14%)
Entorhinal cortex	−4% (−7%, 0.5%)	−5% (−9%, −2%)	−5% (−9%, −2%)	−13% (−21%, −4%)	3% (−4%, 12%)	6% (−3%, 15%)
Hippocampus	−4% (−7%, −1%)	−5% (−8%, −2%)	−5% (−8%, −1%)	−12% (−19%, −5%)	1% (−7%, 9%)	4% (−4%, 13%)
Temporal inf	1% (−2%, 4%)	−11% (−16%, −7%)	−12% (−17%, −8%)	−4% (−12%, 4%)	−1% (−11%, 10%)	−1% (−11%, 11%)
Frontal mid	−4% (−10%, 2%)	−18% (−24%, −13%)	−19% (−25%, −14%)	−1% (−14%, 13%)	−6% (−18%, 7%)	−7% (−19%, 7%)
Dementia
Amygdala	−1% (−6%, 4%)	−5% (−9%, −1%)	−5% (−9%, −1%)	−4% (−14%, 6%)	4% (−4%, 12%)	5% (−4%, 15%)
Entorhinal cortex	−6% (−11%, −0.3%)	−5% (−9%, −1%)	−4% (−9%, −0.2%)	−15% (−24%, −4%)	3% (−5%, 12%)	5% (−4%, 15%)
Hippocampus	−6% (−11%, −2%)	−5% (−9%, −1%)	−4% (−9%, −0.2%)	−12% (−21%, −3%)	1% (−7%, 10%)	4% (−5%, 14%)
Temporal inf	2% (−3%, 6%)	−11% (−16%, −5%)	−12% (−18%, −6%)	−8% (−17%, 2%)	2% (−9%, 15%)	3% (−9%, 17%)
Frontal mid	−2% (−10%, 6%)	−22% (−29%, −14%)	−23% (−31%, −15%)	−3% (−20%, 15%)	−1% (−16%, 15%)	−1% (−16%, 16%)

Data are shown as estimates (95% confidence intervals). Results of multivariate regression model are shown. The upper rows show results from the primary analyses with all participants. The lower rows show results from the sensitivity analyses with subgroup of participants with dementia. Abbreviations: frontal mid = middle frontal cortex; temporal inf = inferior temporal cortex

For the amygdala, age associated with flortaucipir uptake and but not with volume, that a 10-year increase in age correlated with less (−5.2%[95%CI:−8.2%,−2.1%], p=0.001) flortaucipir retention. For the ERC, older participants trended to have lower volumes (−3.5%[95%CI:−7.4%,0.5%],p=0.08) and had significantly lower uptakes (−5.5%[95%CI:−8.6%,−2.3%], p=0.001). Similarly for the hippocampus, older participants showed changes of −4.1%(95%CI:−7.2,−0.9,p=0.01) in volume and −4.9%(95%CI: −8.0%,−1.8%,p=0.003) in uptake every 10 years.

Involvement of neocortical areas were limited to the flortaucipir uptake dimension, with older participants having considerably greater decreases in uptake (−11.4%[95%CI:−15.7%,−7.2%] for inferior temporal and −18.3%[95%CI:−23.7%,−13.0%] for middle frontal; p<0.001 for both).

3.4. TDP-43 effect on volume and flortaucipir uptake

Effect sizes for TDP-43 pathology are shown in Figure 3B and Table 2. Differences were noted in the medial temporal lobe, where TDP-43(+) participants had significantly smaller ERC (−12.7%[95%CI:−21.2%,−3.6%],p=0.007) and hippocampal (−11.8%[95%CI:−18.6%,−4.6%],p=0.002) volumes than TDP-43(−) participants. There were no significant differences between the two groups in terms of volume of the amygdala, inferior temporal cortex, or middle frontal cortex.

In all regions, TDP-43 status was unrelated to flortaucipir uptake after adjusting for Braak NFT stage. Analyses using data with partial volume correction yielded very similar results.

3.5. Sensitivity analyses

Sensitivity analyses including only participants with an antemortem diagnosis of dementia revealed concordant results regarding both volume and uptake changes, indicating that cognitive state had little to no influence on the primary analyses (Figure 2–3, Table 2).

3.6. TDP-43 effect across the age spectrum

The estimated mean volumes and flortaucipir SUVRs by age at scan in the whole cohort, as well as in the TDP-43(+), TDP-43(−), and dementia subgroups are shown in Figure 4. As previously stated, an interaction term between age at scan and TDP-43 status initially included in our model revealed that the TDP-43 effect sizes were not age-dependent, i.e., the TDP-43 effect sizes were largely consistent across the age at scan spectrum. Indeed, the approximately 12–13% differences in hippocampal and ERC volumes between the TDP-43(+) and TDP-43(−) subgroups were maintained throughout the age spectrum, as their mean volumes formed parallel lines in Figure 4A. However, at older ages, the mean volumes of the whole cohort and, more evidently, of the subset with dementia moved closer towards the mean of the TDP-43(+) subgroup. This coincidentally occurred in the presence of increasing frequency of TDP-43(+) participants at older ages. A similar pattern can be seen in the amygdala, inferior temporal and middle frontal cortices, although TDP-43 effect on volume did not reach significance. Comparably, the mean uptakes for the whole cohort also inclined towards the mean of those TDP-43(+) participants in all regions (Figure 4B).

Figure 4. Relationships between mean volumes and flortaucipir SUVRs across the age spectrum.

Regression lines for the whole cohort (gray continuous) and for the TDP-43(+) (pink), TDP-43(−) (green), and dementia subgroups (gray dashed) show the pattern of changes in volume (A) and flortaucipir uptake (B) across the age at scan spectrum. Individual patient points are shown to demonstrate the distribution of data based on TDP-43 status (pink circles for TDP-43(+) and green circles for TDP-43(−)). Abbreviations: Frontal mid = middle frontal cortex; ROI = region of interest; SUVR = standardized uptake value ratio; Temporal inf = inferior temporal cortex

4. DISCUSSION

In this study, we aimed to better characterize the effect of TDP-43 pathology and age on volume and flortaucipir uptake in medial temporal lobe and neocortical regions in an elderly cohort with ADNC. We found that increasing age was associated with lower medial temporal lobe (hippocampus and, to a lesser extent, ERC) volume and lower flortaucipir uptake in all assessed regions. Contrarily, TDP-43 pathology only associated with lower volumes of the hippocampus and ERC and was unrelated to flortaucipir retention across regions. The effect sizes of TDP-43 pathology also appeared to be age-independent, as they remained similar across the age spectrum. However, at older ages, the mean volumes of the cohort generally inclined towards the means of the TDP-43(+) subgroup. This is likely attributable to the higher frequency of TDP-43(+) participants aged $\ge$75 years. The combined presence of TDP-43 pathology effect on hippocampal and ERC volumes, which appeared intensified at older ages due to its increasing prevalence, and the absence of effect on flortaucipir uptake suggest that TDP-43 pathology is a relevant driver of the mismatch between greater volume loss and relatively lower flortaucipir uptake in the medial temporal lobe in older AD participants.

Degeneration of the medial temporal lobe, specifically the hippocampus and ERC, has been widely reported in AD, vascular dementia, TDP-43 proteinopathy, and normal aging.^19,40–42 Medial temporal lobe age-related atrophy results from other age-dependent disease processes like systemic hypertension and chiefly involves the hippocampus, where considerable annual percentage decrease in volume occurs.⁴¹ Age-related decreases in ERC volume has been reported to be minimal and occurring at lower rates.^41,43,44 Our results are consistent with these previous findings. Nonetheless, some authors have reported a faster shrinkage of the ERC and even faster atrophy of the hippocampus in older individuals.^16,41 We believe TDP-43 pathology is behind such observations and that its relationship with age is very convoluted. In a recent study,⁴⁵ we had found that the frequency of TDP-43 pathology increased linearly with age in a large cohort of older adults with ADNC, from 30% of septuagenarians to 42% of octogenarians and eventually 50% of nonagenarians being TDP-43(+). In this study, most of our TDP-43(+) participants were $\ge$75 years old. Although the relative differences in volumes between the TDP-43(+) and TDP-43(−) participants were maintained throughout the age spectrum, it is logical to expect that the overall effects of TDP-43 pathology would increase given its increasing frequency with age. Indeed, the mean volumes of the cohort for all regions assessed moved closer to the means of those TDP-43(+) with increasing age, reflecting the higher cumulative volume loss driven by the higher percentage of TDP-43(+) among the older participants at this end of the age spectrum. The relationship between age and TDP-43 pathology is further complicated by the type of TDP-43 inclusion, since higher odds of displaying TDP-43 type-α inclusions are seen with old age; additionally, this type has been linked to both higher TDP-43 stages and smaller hippocampal volumes than type-β inclusions.³⁶ TDP-43 type-β is most predominant in the amygdala, where it contributes to volume loss.³⁶ We did not find any significant association between age or TDP-43 status and volume of the amygdala in this study possibly due to the small sample size, since we had previously demonstrated that TDP-43 pathology, especially higher stages with widespread pathology, correlated with reduced volumes of the amygdala and hippocampus using a larger cohort.⁴² Other studies from our group have confirmed this correlation of higher stages with reduced volumes of the hippocampus, temporal and frontal lobes^38,46 and that they were more common with old age.⁴⁵ Nonetheless, we had also previously found that even amygdala-only TDP-43 pathology associated with reduced volumes of the ERC, hippocampus and temporal pole possibly due to degeneration of reciprocal projections.⁴⁷ Because of this association between TDP-43 stage and volume, we performed auxiliary analyses on this cohort (see Supporting Information) and found that stage 2+ participants had smaller hippocampal volumes than stage 1 and TDP-43(−) participants, which is in line with our earlier reports.^42,47

After adjusting for Braak NFT stage, age but not TDP-43 pathology had an effect on flortaucipir retention. Focal medial temporal lobe retention is common in old age⁴⁸ while higher neocortical uptake is more frequent in younger AD patients independent of the clinical phenotype.¹⁶ We found that while 10-year age increments only associated with small reductions in flortaucipir retention within medial temporal areas, greater reductions were seen in cortical regions (frontal>temporal), which is in line with previous studies.^16,48 The lack of TDP-43 effect on flortaucipir uptake is also concordant with an autoradiographic study which demonstrated lack of flortaucipir binding to TDP-43 inclusions.⁴⁹ Inconsistent flortaucipir binding however has been described in FTLD-TDP, though the exact nature of this retention remains elusive.¹⁵ Nevertheless, we cannot exclude an indirect effect of TDP-43 pathology on flortaucipir uptake. Flortaucipir binds with higher affinity to NFTs than neuritic plaques and even binds with greater affinity to mature paired helical filaments compared to early pretangles or extracellular “ghost” tangles.¹⁵ In AD, intracellular mature tangles becomes extracellular ghost tangles after neuronal death.⁵⁰ Following the logic that TDP-43 pathology frequency increases with age and causes more volume loss, increased neuronal cell death and ghost tangles will be seen. It is then reasonable to theorize a possible additional, indirect effect of TDP-43 pathology on flortaucipir uptake through the formation of tangle species with less affinity with the radiotracer. However, this remains a hypothesis since our analyses did not show significant associations between TDP-43 pathology and tau uptake.

When concerning AD patients older than 70, it might be prudent to consider TDP-43 co-pathology as the rule rather than an exception. Future studies could investigate the usefulness of said volume-uptake mismatch to predict TDP-43 pathology in older adults with positive AD biomarkers. Clinical trials aimed at finding treatment for AD particularly with the use of anti-tau drugs should be cautious of the effects of TDP-43 pathology when interpreting outcome measures, as these drugs might be less effective among the oldest old. Development of treatment strategies targeting TDP-43 proteinopathy, whether individually or combined with tau, should be pursued. Finally, studies should investigate whether the effectiveness of symptomatic treatments for AD would vary depending on age and TDP-43 status.

A strength of this study is the relatively high number of participants who underwent standardized flortaucipir-PET and neuropathologic protocols. We are aware of the possibility that off-target binding of flortaucipir to the choroid plexus may spill-in to the adjacent hippocampus affecting true uptake measurements.²⁶ For this reason, we decided to include other medial temporal lobe regions, like the amygdala and ERC, which are less affected by this phenomenon. Another limitation is that we only sampled one hemisphere; hence, it is possible that some cases designated as TDP-43(−) could have had TDP-43-immunoreactive inclusions in the other hemisphere. Finally, due to the limitations of statistical models and observational data, we cannot fully delineate the interrelationship between the effects of TDP-43 and aging on volume loss. While undoubtedly other processes are at play, within the limits of our data, our conclusion is that TDP-43 pathology is an important component of volume loss seen at older ages in the medial temporal region.

In summary, the relationship of TDP-43 pathology with volume and flortaucipir uptake across the age spectrum is not straightforward. Old age is associated with lower medial temporal lobe volumes and lower flortaucipir uptake in both medial temporal lobe and cortical areas. Contrarily, TDP-43 pathology is independently associated with decreased volumes of the hippocampus and ERC alone. However, although the amount of TDP-43 effect remains constant across the age spectrum, its overall effect on volume among the oldest old is magnified because of the cumulative effects driven by the increasing percentage of TDP-43(+) in this population. While it is likely that several factors other than TDP-43 pathology are acting synergistically, our data show that TDP-43 pathology is a relevant contributor of the mismatch between greater lower medial temporal volumes despite relatively lower flortaucipir uptake in older adults with AD. Research on the pathogenesis and development of treatment strategies for AD should factor in the increasing frequency and effect of TDP-43 pathology in older age.

RESEARCH IN CONTEXT.

Systematic review:

The authors reviewed the literature from PubMed regarding the different patterns of medial temporal volume loss and tau radiotracer uptakes across AD phenotypes. While distinct patterns are seen between the old- and young-onset AD, little is known about the etiology of the mismatch between greater medial temporal volume loss despite less tau uptake in AD patients $\ge$70 years.

Interpretation:

This study contributes to the knowledge about how TDP−43 pathology is a key factor influencing neurodegeneration, as seen with regional changes in volume and tau uptake, particularly in older AD patients.

Future directions:

These findings highlight the importance of TDP−43 co-pathology in AD. Future studies could investigate the usefulness of the volume-uptake mismatch in predicting TDP−43 co-pathology. Clinical trials aiming to find treatment for AD should consider the effects of TDP−43 pathology when interpreting outcome measures. Development of treatment strategies targeting TDP−43 pathology in AD should also be pursued.

HIGHLIGHTS.

• TDP−43 pathology affects medial temporal volume loss but not tau radiotracer uptake

• Greater TDP−43 pathology effect is seen in old age due to its increasing frequency

• TDP−43 pathology is a relevant driver of the volume-uptake mismatch in old AD patients