Development and Clinical Validation of Global Tau Severity Score in Young- and Late- Onset Alzheimer's Disease Using Florzolotau (18F) PET

doi:10.21203/rs.3.rs-3932300/v1

Download PDF

Research Article

Development and Clinical Validation of Global Tau Severity Score in Young- and Late- Onset Alzheimer's Disease Using Florzolotau (18F) PET

https://doi.org/10.21203/rs.3.rs-3932300/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Background

Tau-specific positron emission tomography (tau-PET) holds promise in distinguishing Alzheimer's disease (AD) and elucidating the spatial distribution of tau deposition. In particular, the phenotypic differences between the young-onset AD [YOAD] and late-onset AD [LOAD] likely arise from the distinct biological properties of tau proteins to their downstream pathways. This study aimed to establish a global tau severity (gTS) scale based on Florzolotau (18F) PET, a highly specific second-generation tau tracer for diagnosing AD, for standardizing tau burden quantification.

Methods

A total of 186 participants were enrolled and divided into a pilot group (15 cognitive unimpaired controls [CTLs] and 15 AD patients) and a validation group (27 CTLs, 67 patients with YOAD, and 62 patients with LOAD). The pilot group results were utilized to create an AD-specific tau mask and determine the optimal Florzolotau (18F) reference region based on effect size. In the validation group, cutoffs for diagnosing YOAD and LOAD using the gTS score were calculated. Regression models were used to assess the impact of amyloid centiloid, gTS score, and hippocampal volume on cognitive outcomes.

Results

The white matter region was determined to be the most suitable reference for Florzolotau (18F). The gTS cutoff values of 24.1 for both AD and YOAD and 34.1 for LOAD demonstrated highest diagnostic accuracy, as indicated by the area under the curve. The gTS scores significantly predicted total scores and subdomains on cognitive ability screening instruments. Cognitive-gTS curve features were found to have quadratic and linear relationships with YOAD and LOAD, respectively, illustrating the direct effect of tau pathology on cognition.

Conclusions

The gTS score, derived from Florzolotau (18F) PET scans, provides a robust method for assessing global tau burden. The scale reveals different cognition–tau relationships in YOAD and LOAD, indicating distinct pathological property of tau on disease progression.

Alzheimer's disease

cognition

early onset

Florzolotau (18F)

global tau severity score

In 2018, the National Institute on Aging-Alzheimer's Association (NIA-AA) established a research framework for Alzheimer's disease (AD) [1]. In this framework, amyloid and tau positron emission tomography (PET) images are used for the in vivo visualization of pathological changes[2–6]. Moreover, researchers found a strong correlation between cognition test scores and the spatial spread of tau PET signals [7–11]. To address the problem of off-target binding, a second-generation tau tracer, Florzolotau (18F), has been developed [12–14]. Florzolotau (18F) demonstrated high in vitro binding specificity, as confirmed through histopathology staining [15]. Methods for quantifying Florzolotau (18F) PET image signals include those based on the regional standard uptake value ratio (SUVr); [16] and visual rating scales [17]. Although results from these methods are strongly correlated with clinical observations, a tau-based composite score for quantifying global tau burden remains to be developed. Ideally, the tau composite score would assist in AD diagnosis by offering high classification accuracy and mirroring salient cognitive measurements.

AD manifests in two primary age groups: late-onset AD (LOAD), which occurs in individuals over 65-years old, and young-onset AD (YOAD), which accounts for 5.5% of all AD cases [18]. Both YOAD and LOAD exhibit pathological burdens in the form of amyloid-β plaques and tau tangles, but they differ phenotypically and genetically. Patients with YOAD commonly exhibit less impairment in episodic and semantic memory [19] and tend to be receive a diagnosis at more advanced clinical stages with genetic predispositions [20]. Additionally, these patients experience greater impairments in visuospatial, executive, and linguistic functioning. Conversely, LOAD is a highly polygenic condition, with known risk variants accounting for approximately 8% of variances in genome-wide studies [21]. Patients with LOAD experience a slower cognitive decline and exhibit a higher prevalence of age-related copathologies [22–24]. A reliable biomarker to reflect the disease progression pattern in these two age groups is fundamental. In particular, the phenotypic differences between the two age groups likely arise from the distinct biological properties of tau proteins to their downstream pathways.

The study aimed to develop and validate a global tau severity (gTS) score using Florzolotau (18F) PET images. The gTS score was initially formulated based on the unit of centiloid in a pilot cohort comprising 15 cognitive unimpaired controls (CTLs) and 15 patients with AD. The analytical approach for constructing the score involved determining the optimal reference region for standard uptake value ratio (SUVr) calculations and selecting the most suitable AD-specific volume of interest (VOI) for Florzolotau (18F). Subsequently, we applied the gTS score formula to another independent cohort (n = 156). Validation was conducted by comparing the score with two independent raters’ read out. For clinical applications, we determined the optimal cutoff values for diagnosing YOAD (n = 67) and LOAD (n = 62) using a reference group of 27 CTL participants. Next, we assessed the predictive power of the gTS score for cognitive outcomes. For this assessment, we used a forced-entry multiple regression analysis model with independent variables, such as amyloid centiloid unit, gTS score, and hippocampal volume. According to tau propagation theory [25], tau proteins exhibit prion-like characteristics, and a specific subset of these proteins initiates the degeneration of gray matter (GM). The ultimate goal was to understand whether gTS score reflected global tau burden and, if so, the influence of the global tau burden on cognitive outcomes should be direct and not mediated indirectly by the hippocampal volume.

Participants

This study was conducted in accordance with the Declaration of Helsinki and was approved by the Institutional Review Board of Chang Gung Memorial Hospital. All individuals in this study were recruited from the Taiwan-ADNI (https://tadni.cgmh-mi.com) or the cognition and aging cohort, both were ongoing, multicenter, hospital-based, longitudinal cohorts with established clinical protocols [26] [27, 28].

Group stratification and diagnosis

Beginning in August 2019, the study subsequently enrolled participants meeting the criteria for three groups: CTL, YOAD, and LOAD. In total, 186 individuals (42 CTL and 144 AD) completed the study. The pilot cohort (15 CTL and 15 AD) included patients from the Chang Gung Memorial Hospitals of Taipei and Linkou. The validation cohort (n:156, 27 CTL, 67 YOAD, and 62 LOAD) included patients from the Chang Gung Memorial Hospitals of Kaohsiung and Feng Shang. The tau image scores were separately rated by two independent raters with 1 = no uptake, 2 = equivocal, 3 = mild uptake, and 4 = moderate to severe uptake[17].

Cognitively unimpaired CTL participants were referred by the neurologists, defined by the absence of cognitive complaints, a clinical dementia rating (CDR) of 0 [29], and a mini-mental state examination (MMSE) score ≥ 24. All the CTL had an amyloid centiloid level < 10.

Both YOAD and LOAD participants fulfilled the NIA-AA 2018 clinical diagnostic criteria[1], a MMSE score < 24 and clinical dementia rating ≥ 0.5, exhibited positive visual read outs of amyloid PET by two independent raters. All the PET images of AD had amyloid centiloid score > 30.

The exclusion criteria included a history of clinical stroke, a modified Hachinski ischemic score of > 4, a degenerative brain disease other than AD, lesions on T2-weighted magnetic resonance imaging (MRI) indicative of severe white matter (WM) diseases, clinically unmanaged diabetes, and diagnoses of major depressive disorder or dysthymic disorder according to the Diagnostic and Statistical Manual of Mental Disorders, fourth edition, text revision (DSM-IV-TR).

Cognitive assessment

A trained neuropsychologist administered a battery of neurobehavioral tests to assess cognitive function, including the MMSE and cognitive ability screening instrument (CASI). The CASI is divided into nine subdomains, with executive domains comprising attention, verbal fluency, abstract thinking, and mental manipulation [30] and nonexecutive domains comprising orientation, short-term memory (STM), long-term memory, language ability, and visual construction [30].

Data acquisition and image reconstruction

Florzolotau (18F) PET, F¹⁸-Florbetapir PET, and 3D T1 weighted MRI scans were obtained from all participants. 3D T1-weighted magnetization prepared rapid gradient echo (MP-RAGE) sequencing was performed using a 3T Siemens machine equipped with a 32-channel phased array head coil. The acquisition parameters were as follows: a repetition time of 2600 ms, an echo time of 3.15 ms, a matrix of 256 × 224, 176 continuous slices in the sagittal plane, and an in-plane spatial resolution of 0.5×0.5×1 mm³. This MP-RAGE scan was used for partial volume correction in PET imaging and enabled volumetric comparisons among biomarker groups.

Both F¹⁸-Florbetapir and Florzolotau (18F) were synthesized at the cyclotron facility of Chang Gung Memorial Hospital. PET scans were acquired using a GE Discovery MI PET/CT scanner. For F¹⁸-Florbetapir, the acquisition protocol, optimal scanning time, image reconstruction, and amyloid Centiloid scores followed established guidelines [28, 31, 32].

Brain tau PET scans were performed 90 min after injection of 185 ± 74 MBq of Florzolotau (18F). The acquisition window lasted 10 min and comprised two 5-min dynamic frames for motion correction. These PET images underwent motion correction and were then aggregated into a static frame image in their native space. Subsequently, spatial normalization to the Montreal Neurological Institute space was performed using an MRI-based spatial normalization method. The 3D PET images were acquired and reconstructed using an iterative reconstruction algorithm (OSEM, four iterations and 16 subsets) and further processed with a post hoc 5 mm Gaussian filter. For attenuation correction, low-dose CT scans were employed with parameters set at 15 mAs, 120 keV, 512 × 512 matrix, a 2.79-mm slice thickness, 71 slices, 110 mm/s increment, a 0.5-s rotation time, and a pitch of 1.375.

Pilot data: gTS score formula

To identify the AD-specific VOI that retains Florzolotau (18F), we first selected 15 patients with moderate to severe AD (CDR > 1; 70.1 ± 5.7 years; range 58–75 years, 6 males) and 15 CTL (68.2 ± 7.6 years; range 54–84 years, 8 males). Although no significant age differences existed between the CTL and AD groups, the patients with AD had significantly higher amyloid centiloid scores compared with CTL (AD: 129.59 ± 14.3; CTL: −9.41 ± 5.69). Voxel-based morphometry analyses were conducted on Florzolotau (18F) images, with age and estimated total intracranial volume (eTIV) used as nuisance covariates. A height threshold of p < 0.0001 was set, with family-wise error corrections and an extent threshold of 200 voxels. This identified region was used as the AD-specific neocortical VOI for calculating the gTS Centiloid score.

To determine the optimal reference region for SUVr, we evaluated five reference regions. First, in the WM region, the mean value and the parametric estimation of reference signal intensity (PERSI) method were calculated separately. The PERSI method was calculated for each scan by fitting a bimodal Gaussian distribution to the voxel-intensity histogram within an atlas-based WM region. The center and width of the lower-intensity peak were then used to identify the voxel intensities included in the calculation[33]. Additionally, we examined three reference regions from the GAAIN website (http://www.gaain.org/centiloid-project); these regions were the whole cerebellum, cerebellar GM, and a combination of the whole cerebellum and brainstem.

We then calculated the AD-specific neocortical VOI values using the SUVr images. We established a zero-anchor point based on the average neocortical VOI values of the 15 CTL participants and defined a 100-anchor point using the average neocortical VOI values of the 15 patients with AD.

The mean SUVr values were converted to Centiloid (CL) values using the following standard equation:

$$\text{CL}=\frac{\text{SUVr}\text{IND}- \text{SUVr}\text{CN-0}}{\text{SUVr}\text{AD-100} - \text{SUVr}\text{CN-0}}\times 100$$

where SUVr _IND is an individual’s SUVr VOI value, SUVr _CTL−0 is the mean SUVr VOI value of the CTL group, and SUVR _AD−100 is the mean SUVr VOI value of the AD group. The CL represents the gTS score in this study.

Hippocampal volume and eTIV

Hippocampal volume was measured using automated segmentation from 3D T1-weighted images, employing the FIRST tool available in the Oxford Centre for Functional MRI of the Brain Software Library (FSL) 5.0 and adhering to the ENIGMA1 protocol [34]. The segmentation process involved reviewing images at key intermediate points, including after registration and following skull stripping, and making manual adjustments when necessary. The eTIV was calculated using the ENIGMA1 protocol for FSL. This process required the linear alignment of each participant’s brain with the Montreal Neurological Institute 152 (MNI152) template. The inverse of the determinant of the affine transformation matrix was then multiplied by the template size to yield a robust eTIV estimate [35]. All affine transformations were visually inspected and adjusted, ensuring their accuracy. In this study, the hippocampal volume ratio (%) was defined as the fraction of the right hippocampal volume relative to the eTIV, formulated as (right hippocampal volume / eTIV) × 100. This choice was motivated by the observation that AD typically manifests symmetric neurodegeneration.

Statistical analysis

Clinical data were expressed as mean and standard deviation. We tested the normality of the data using the Shapiro–Wilk test. Analysis of variance was used for comparisons of continuous variables between the YOAD, LOAD, and CTL groups. Bonferroni corrections were applied for multiple comparisons. Correlation analysis was performed using Spearman’s correlation analysis, and adjustments were made for possible confounders as detailed. Statistical significance was set at p < 0.05. The statistical analyses were performed using R software version 4.2.1.

To compare the performance of the five reference regions for SUVr calculations, we analyzed the coefficient of variance, and the effect sizes of the pilot group differences. The region with the highest effect size was identified as the most suitable reference region. Subsequently, we extracted Braak’s neurofibrillary tangle (NFT) signals of interest in vivo based on this optimal reference region for SUVr [36–38].

For validation, we applied the gTS score formula to another independent cohort. Distributions of gTS scores and the visual read out [17] from two independents raters were compared.

We used receiver operating characteristic (ROC) curve analysis, using the highest area under curve (AUC) as a metric to detect the cutoff values to diagnose YOAD or LOAD with the CTL as the reference group [39]. The ROC curves were generated to visualize the classification accuracy, wherein sensitivity (y-axis) was plotted against 1 − specificity (x-axis). The optimal cut-points were determined using the Youden index (sensitivity + specificity − 1).

To investigate the predictive power of gTS score on cognitive performance, we conducted forced-entry multiple regression analysis. In these models, the dependent variables were the CASI total score, its executive or nonexecutive domains, or its nine subdomains. The predictor variables in the regression models were the gTS value, amyloid Centiloid score, hippocampal volume, age, education, and gender. Consequently, a total of 24 regression models were computed for the YOAD and LOAD. To account for multiple comparisons, the significance level (α) was corrected using a Bonferroni factor of 24 (α = 0.05/24 = 0.00208). Subsequently, linear and quadratic models were fitted to examine the relationship between tau Centiloid score and cognition in the YOAD and LOAD cohorts. A linear model was selected if the quadratic parameter of a quadratic model was not significant.

To understand whether the gTS score can serve as a direct cognitive marker or whether its relationship with cognition is mediated by another indirect pathway such as hippocampal volume integrity, we further performed mediation analysis [40] where the gTS score served as the predictor, the hippocampal fraction served as the mediator, and the cognitive test scores served as the outcomes. We used bootstrapping tests with 1000 resamples and bias-corrected confidence intervals to establish the mediation model. In our model, the total direct pathway was gTS score → cognition and the indirect pathway was gTS score → hippocampal volume → cognition. For two AD groups, we used the CASI total score, executive, nonexecutive, and STM score as the cognitive outcomes. A significance level of p < 0.05 was established for this analysis.

Pilot sample and gTS score formula

Supplementary Fig. 1 illustrates the average distribution of Florzolotau (18F) in the CTL group (supplementary Fig. 1A) and the AD group (supplementary Fig. 1B). Notable clusters with higher Florzolotau (18F) levels in the AD group were observed in the precuneus, medial temporal, and bilateral frontal regions (supplementary Fig. 1C). These areas were binarized and converted to form an AD-specific neocortical mask (supplementary Fig. 1D).

Supplementary Table 1 provides the SUVr of Florzolotau (18F), calculated using five different reference methods in the pilot sample. Among these, the WM region yielded the highest effect size, and the mean value in the WM region had higher effect size than the PERSI method. Therefore, we chose the WM mean value as the optimal reference method for Florzolotau (18F) PET imaging calculations.

The Centiloid (CL) equation for calculating the gTS score for Florzolotau (18F) is as follows:

$$gTS score=\frac{\text{SUVr}\text{IND}-0.907 }{0.415 }\times 100$$

Demographic characteristics of validation dataset

Demographic data of the validation dataset is presented in Table 1. The LOAD group contained a higher proportion of women, was older, and had a lower eTIV compared with the YOAD group.

Table 1

Demographics data characteristics
	CTL (n = 27)	AD (n = 129)	CTL v.s AD T (p value)	AD		CTL, YOAD, LOAD F (p value)	Post-hoc Analysis
	CTL (n = 27)	AD (n = 129)	CTL v.s AD T (p value)	YOAD (n = 67)	LOAD (n = 62)	CTL, YOAD, LOAD F (p value)	Post-hoc Analysis
Age [years]	67.0 (7.7)	70.4 (5.8) ^*	-2.634 (0.009)	66.5 (4.2)	74.6 (4.1) ^*	48.543 (< 0.001)	LOAD > YOAD = CTL
Education[years]	10.5 (4.6)	9.6 (4.3)	1.038 (0.301)	10.4 (4.2)	8.7 (4.3)	3.036 (0.051)	-
Female (%)	12 (44.4%)	80 (62.0%)	0.131	34 (50.7%)	46 (74.2%) ^*	0.006	LOAD > YOAD = CTL
Disease duration (years)	N.A.	4.3 (2.6)		4.4 (2.6)	4.1 (2.6)	0.314 (0.576)
Mini-Mental State Examination	27.7 (2.4)	20.4 (5.9) ^*	10.365 (< 0.001)	20.6 (6.6) ^*	20.2 (5.1) ^*	19.579 (< 0.001)	CTL > YOAD = LOAD
Clinical dementia rating (CDR)	0	0.7 (0.4) ^*	-20.813 (< 0.001)	0.7 (0.4) ^*	0.7 (0.4) ^*	45.097 (< 0.001)	CTL > YOAD = LOAD
CDR_Sum of box	0	3.6 (2.8) ^*	-14.539 (< 0.001)	3.8 (3.0) ^*	3.3 (2.6) ^*	22.680 (< 0.001)	CTL > YOAD = LOAD
CASI total scores (100)	91.1(6.4)	69.8 (18.4) ^*	10.385 (< 0.001)	70.7 (20.4) ^*	69.0 (16.2) ^*	17.449 (< 0.001)	CTL > YOAD = LOAD
CASI subdomains
Mental Manipulation(10)	9.1(1.1)	7.3(2.8) ^*	5.767 (< 0.001)	7.7 (2.8) ^*	6.9 (2.8) ^*	7.360 (< 0.001)	CTL > YOAD = LOAD
Attention (8)	7.5 (0.6)	6.4 (1.4) ^*	6.398 (< 0.001)	6.5 (1.4) ^*	6.3 (1.3) ^*	9.078 (< 0.001)	CTL > YOAD = LOAD
Orientation (18)	17.2 (1.8)	11.6 (5.2) ^*	9.740 (< 0.001)	12.0 (5.2) ^*	11.2 (5.1) ^*	15.618 (< 0.001)	CTL > YOAD = LOAD
Long Term Memory(10)	10.0 (0)	9.0 (2.2) ^*	5.147 (< 0.001)	8.8(2.5) ^*	9.2 (1.9)	3.564 (0.031)	CTL > YOAD
Short Term Memory(12)	10.4 (2.2)	4.2 (3.1) ^*	12.174 (< 0.001)	4.5 (3.4) ^*	4.0 (2.9) ^*	47.253 (< 0.001)	CTL > YOAD = LOAD
Abstract Thinking (12)	9.4 (2.0)	8.0 (2.7) ^*	2.638 (0.009)	7.9 (2.8) ^*	8.1 (2.6)	3.561 (0.031)	CTL > YOAD
Drawing Ability (10)	9.9 (0.6)	8.7 (2.4) ^*	4.920 (< 0.001)	8.7 (2.6)	8.7 (2.1)	3.245 (0.05)	-
Verbal Fluency (10)	7.9 (1.9)	5.7 (2.6) ^*	4.235 (< 0.001)	5.9 (2.8) ^*	5.4 (2.4) ^*	9.732 (< 0.001)	CTL > YOAD = LOAD
Language(10)	9.6 (0.8)	8.9 (1.9) ^*	3.138 (0.002)	8.6 (2.3) ^*	9.3 (1.4)	3.805 (0.024)	CTL > YOAD
CASI executive functional score	34.0 (3.9)	27.4 (7.7) ^*	6.573 (< 0.001)	28.1 (8.4)	26.7 (6.9)	10.166 (< 0.001)	CTL > YOAD = LOAD
CASI non-executive functional score	57.0 (3.5)	42.5 (11.8) ^*	11.794 (< 0.001)	42.6 (13.0)	42.3 (10.5)	19.940 (< 0.001)	CTL > YOAD = LOAD
eTIV(mm³)	1423.9 (179.0)	1388.2 (137.6)	0.979 (0.335)	1420.8 (133.9)	1353.0 (133.8)	4.345 (0.015)	YOAD > LOAD
Right hippocampus volume (mm³)	3.6 (0.4)	2.7 (0.7) ^*	8.291 (< 0.001)	2.8 (0.7) ^*	2.6 (0.6) ^*	22.314 (< 0.001)	CTL > YOAD = LOAD
Right hippocampus volume/ eTIV ratio (%)	0.251(0.027)	0.195 (0.043) ^*	8.672 (< 0.001)	0.196 (0.042) ^*	0.193 (0.044) ^*	21.146 (< 0.001)	CTL > YOAD = LOAD
Tau Centiloid Score	2.3 (16.3)	85.4 (38.1) ^*	-18.093 (< 0.001)	84.7 (40.9) ^*	86.3 (35.2) ^*	61.164 (< 0.001)	CTL < YOAD = LOAD
Amyloid Centiloid Score	2.7 (10.0)	74.6 (29.9) ^*	-23.716 (< 0.001)	73.9 (28.7) ^*	75.3 (31.3) ^*	87.431 (< 0.001)	CTL < YOAD = LOAD

Data are means (standard deviation) or number (%). * p < 0.05 in comparison with CTL;

AD, Alzheimer disease; CTL, cognitively normal subjects; YOAD: young-onset Alzheimer disease; LOAD: late-onset Alzheimer disease;

CASI: Cognitive Abilities Screening Instrument; eTIV: estimated Total Intracranial Volume.

The CASI executive functional score was sum of attention, verbal fluency, abstract thinking, and mental manipulation scores. The orientation, short- and long-term memory, language ability, and drawing were regarded as nonexecutive functional score.

The gTS and amyloid scores distribution showed a clear cluster boundary between AD and CTL (Fig. 1A). According to the gTS distributions, the classification between scores 1 and 2 in one rater was not significant (Fig. 1B).

Significant differences in tau topography between the LOAD and CTL (Fig. 2A) as well as between the YOAD and CTL groups (Fig. 2B) were observed. The increased Florzolotau (18F) signals in these groups were primarily localized in the hippocampus, fusiform gyrus, parahippocampal gyrus, amygdala, dorsolateral prefrontal lobe, lateral temporal cortex, inferior temporal gyrus, supramarginal gyrus, angular gyrus, posterior cingulate gyrus, precuneus, and occipital cortex. However, no discernible differences in tau PET topography between the YOAD and LOAD groups were found. The CTL group did not exhibit any age-related decline in gTS scores.

The gTS score and Braak’s NFT signals of interest

Figure 2C displays the gTS scores across Braak stages of the YOAD, LOAD, and CTL groups for comparison. The gTS scores revealed similar patterns of variation between YOAD and CTL and between LOAD and CTL, with no significant differences observed between YOAD and LOAD. In the YOAD group, the gTS score was significantly correlated with Braak’s stages I–II (ρ = 0.3, p = 0.01), III–IV (ρ = 0.76, p = 3.18 x 10^− 14), and V–VI (ρ = 0.73, p = 1.16 x 10^− 12). In the LOAD group, significant correlations were only noted for stage III–IV (ρ = 0.62, p = 1.48 x 10^− 7) and V–VI (ρ = 0.58, p = 1.16 x 10^− 7). Specifically, in the YOAD group, age at the time of tau PET imaging inversely correlated with Braak stages III–IV and V–VI (Fig. 2D). No such correlation between age and Braak stage was observed in the LOAD group.

Optimal gTS cutoff value for diagnosing YOAD or LOAD

A gTS score of 24.1 was the optimal diagnostic cutoff for distinguishing between the AD and CTL groups (Supplementary Fig. 2A), with an AUC of 0.969, a sensitivity of 94.6%, and a specificity of 92.6%. This cutoff could also be used for diagnosis in the YOAD group (Supplementary Fig. 2B), with an AUC of 0.956, a sensitivity of 92.5%, and a specificity of 92.6%. For the LOAD group (Supplementary Fig. 2C), a slightly higher cutoff value of 34.1 yielded an AUC of 0.984, a sensitivity of 93.5%, and a specificity of 96.3%. No optimal cutoff emerged for differentiating between the LOAD and YOAD groups (Supplementary Fig. 2D).

The gTS score offers robust cognition prediction

The regression analysis results are detailed in Supplementary Table 2–13. For the LOAD and YOAD groups, we found the gTS score, educational year, and hippocampal volume were significantly predictive of CASI total scores (Supplementary Table 2). In the LOAD and YOAD groups, the gTS score was predictive of executive or nonexecutive domains (Supplementary Table 3,4), abstract thinking scores, (Supplementary Table 7) and STM scores (Supplementary Table 11). Across all the predictive models, neither amyloid Centiloid score nor gender made a significant contribution to predicting cognitive outcomes.

Curve fitting differences in YOAD and LOAD

To explore the similarity and differences between LOAD and YOAD in salient cognitive features, we employed curve fitting models using the gTS score to simulate tau levels. These models targeted CASI total scores (Fig. 3A), STM (Fig. 3B), executive functions (Fig. 3C) and nonexecutive functions (Fig. 3D). For the LOAD group, the optimal curve fitting model was a linear regression model. For the YOAD group, a linear curve was predictive of STM, whereas quadratic curves were more predictive of the CASI total, nonexecutive, and executive domain scores.

The hippocampal fraction-cognition fitting model aligned with the findings from the gTS model for CASI total scores (Fig. 3E). In the YOAD group, linear regressions were observed for STM, executive, and nonexecutive domains (Fig. 3F, 3G, 3H). Conversely, in the LOAD group, a linear model fit data on executive function well, whereas quadratic curve patterns fit data on STM and nonexecutive domains well.

Mediation analysis

Mediation analysis was performed for both the LOAD and YOAD groups to explore whether gTS scores could serve as a direct cognitive biomarker (Fig. 4). For both the YOAD and LOAD groups, all predefined cognitive outcomes were directly influenced by the gTS score. The differences between two AD groups emerged only in the executive domain. Specifically, in the YOAD group, the gTS score had a direct effect on executive function without mediation by hippocampal volume. Conversely, in the LOAD group, executive functions were influenced both directly by the gTS score and indirectly through mediation by hippocampal volume.

Major findings

In the context of YOAD and LOAD, our study yielded three major findings. First, we determined optimal gTS score cutoff values for diagnosing YOAD (score = 24.1) and LOAD (score = 34.1), which demonstrated high sensitivity and specificity. These values offer supplementary information that could benefit nuclear medicine specialists during the visual rating procedure especially at the low tau burden stage. Second, the usage of the gTS score in predicting cognitive outcomes was confirmed. Our curve fitting models further clarified the possible differences in the relationships between tau pathology and cognitive performance in YOAD and LOAD. Finally, we found that gTS score serves as a direct cognitive biomarker in CASI total scores, executive, nonexecutive, and STM scores for both the YOAD and LOAD groups. In summary, our study indicates that gTS score cutoff values can be used for diagnosing AD, predicting cognitive scores, simulating the relationships between cognitive and tau burden, and serving as a direct cognitive predictor independent of hippocampal volume.

Optimal cutoff value to complement visual readouts and diagnose AD

For classical AD presenting with cognitive symptoms, amyloid burden often stabilizes when clinical symptoms manifest. Given that amyloid tracers were developed earlier than tau tracers, a greater consensus exists on the visual interpretation of amyloid PET scans [41–43]. Conversely, multiple tau tracers are under development, and inter-rater consistency often depends on standardized preprocessing steps and a user-friendly color scale. Our study provided a proper reference region, tau-related VOI, and calculation formula for Florzolotau (18F). Because our AD-specific regional mask was based on the distribution of Florzolotau (18F) tracer between matched groups, the gTS score formula cannot be directly applied to other tau tracers. The binding topographies reported from major AD cohorts may differ between ethnicities [44, 45].

According to studies on Florzolotau (18F), the WM reference region remains the optimal region, and the quantification using the PERSI method [17, 46] has been validated. Although Tagai et al. reported that the WM reference may exhibit age-related declines [47], such a finding was not observed in our study. The use of mean value in WM was also applicable, and the finding may be more easily applied in clinical use.

Age of onset affect the influence of pathological tau severity to cognition

Studies on the longitudinal propagation of tau indicate distinct spreading pathways between YOAD and LOAD [48]. ADNI longitudinal data reveal faster tau accumulation in patients with YOAD [49], suggesting that the age of onset in AD may directly influence tau spread. Studies have reported that patients with YOAD exhibit greater global tau PET binding and a higher neocortical-to-medial temporal binding ratio compared with patients with LOAD [7, 8]. In our YOAD group, age at the time of tau imaging was inversely related to tau burden in Braak stages III–VI. Although the gTS scores did not differ significantly between LOAD and YOAD in our study population, the YOAD group had a significantly larger eTIV at the time of scanning, which suggests greater brain reserve against tau burden [50].

Age itself may modulate the interactions between tau and Aβ. Tau and Aβ42 are positively related at 45 to 70 years old but negatively and linearly related at 70 years and older [51]. Longitudinal imaging studies suggest that global amyloid accumulation is a prerequisite for the pathologic spread of tau, representing an amyloid-facilitated tauopathy in AD [52, 53]. The age-dependent variations in tau and Aβ interactions may contribute to the differing patterns of tau-cognition relationships observed between YOAD and LOAD.

Furthermore, we found no differences between YOAD and LOAD in terms of fundamental pathological substrates (tau topography, gTS score, signals derived from three Braak’s stages, and amyloid Centiloid), neuronal injury biomarker (hippocampal volume), and cognitive scores. This suggests that both AD groups may exhibit some degree of upstream homogeneity in their pathological characteristics.

Distinct biological properties of tau on cognitive measurement in LOAD and YOAD

Our study results indicate that the gTS score potentially serves for simulating cognitive-tau relationships. Longitudinal follow-up studies are required to understand its applicability for monitoring disease progression. The curve fitting analysis revealed distinct biological properties of tau on cognitive measurements in LOAD and YOAD. Studies have demonstrated that tau propagation is accelerated in the cortical hubs of patients with YOAD [49], leading to earlier tau saturation [54], neuronal degeneration [55], and faster cognitive decline despite a plateau in tau load. Genetic predispositions for YOAD include mutations in amyloid precursor protein, presenilin 1, and presenilin 2 [56], potentially resulting in an overproduction of amyloid plaques. Interactions between tau and Aβ appear to be dependent on the age of onset. Longitudinal imaging studies have suggested that global amyloid accumulation is required for the pathologic spread of tau, representing an amyloid-enabled tauopathy in AD [52].

The age-dependent variation in the interaction between tau and Aβ may account for the observed differences in tau-cognition relationships between YOAD and LOAD. In alignment with these findings, our study revealed a lower diagnostic cutoff value for the gTS score in YOAD compared with LOAD. Furthermore, the curve fitting model suggested a quadratic relationship between the gTS score and cognitive scores in YOAD, whereas a linear relationship was observed in LOAD. Consequently, the YOAD group may manifest clinically at a lower gTS score with earlier saturation, even as cognitive scores continue to decline. Comorbid pathologies contributing to LOAD include vascular damage, TDP-43, and α-synuclein [57]. These factors appear to have less effect than age or the interactive effects of amyloid, which helps to explain the differences between YOAD and LOAD.

Biological properties of tau and the indirect mediation by hippocampus in salient features

Our regression results suggested that both gTS score and hippocampal volume were predictors of CASI total scores and the nonexecutive domain in both AD groups. Moreover, the gTS score alone was predictive of performance in the executive domain, abstract thinking, and STM subdomains across age groups. Studies have established a close correlation between cognitive decline and both tau pathology and hippocampal volume [58–60]. For instance, Visser et al. demonstrated that tau burden was related to STM impairment in YOAD and executive dysfunction in LOAD [8]. Our study extended this literature by suggesting that tau affects a broader range of cognitive domains, some of which are not mediated by hippocampal volume. Conversely, our results illustrated no predictive role of amyloid Centiloid on cognitive performance in either age group, consistent with relevant reviews [61]. Overall, these results suggest that distinct disease pathways may exist in YOAD and LOAD, resulting in differential patterns of tau distribution across different age groups.

Limitations

Several limitations warrant mention. First, the gTS score in this study relied on a cross-sectional design for both its formulation and validation. Therefore, the observed relationships between cognitive performance and gTS scores in YOAD and LOAD groups do not necessarily imply causality in tau propagation. Future research for the validation of the gTS score in a longitudinal cohort is essential, and particular attention should be paid to critical aspects of longitudinal processing, such as temporal effects and the functional-structural co-registration step. Second, our results may have limited generalizability across ethnicities and nationalities because our cohort only comprised Taiwanese people of Han Chinese descent. Nonetheless, the sample was sufficiently large. Third, our selection of WM as the optimal reference region for this study was based on its superior discriminatory power between AD and CTL groups, making it potentially unsuitable for use in other tauopathies. Finally, our curve fitting model used in this study to access the relationship between tau burden and cognitive tests may not fully capture all possible confounders, particularly as vasculopathy and inflammation processes also interact with tau topography. Advances of plasma and imaging biomarkers will likely allow for more nuanced inclusion of factors such as vascular insult, TDP-43, or neuroinflammation in future precision medicine approaches.

The establishment of the Florzolotau (18F) gTS score assist diagnostic accuracy and disease severity monitoring, offering a valuable metric for assessing cognitive performance in both YOAD and LOAD. The influence of tau on downstream cognitive scores differs between YOAD and LOAD, and various confounding factors—such as age-related tau accumulation and coexisting pathologies—contribute to differing degrees. Further longitudinal research is required to investigate the tau propagation circuits in both groups and for clarifying the causal relationships between tau deposition and cognitive decline.

AD	Alzheimer's disease
AUC	area under curve
CASI	cognitive ability screening instrument
CDR	clinical dementia rating
CL	Centiloid
CTLs	cognitive unimpaired controls
DSM-IV-TR	Diagnostic and Statistical Manual of Mental Disorders, fourth edition, text revision
eTIV	estimated total intracranial volume
GM	gray matter
gTS	global tau severity
LOAD	late-onset AD
MMSE	mini-mental state examination
MNI152	Montreal Neurological Institute 152
MP-RAGE	magnetization prepared rapid gradient echo
MRI	magnetic resonance imaging
NFT	neurofibrillary tangle
NIA-AA	National Institute on Aging-Alzheimer's Association
PERSI	parametric estimation of reference signal intensity
ROC	receiver operating characteristic
STM	short-term memory
SUVr	standard uptake value ratio
VOI	volume of interest
WM	white matter
YOAD	young-onset AD

Acknowledgements: The authors gratefully thank all study participants for their participation in this study. This manuscript was edited by Wallace Academic Editing.

Authorship contribution: MNL, CWH, SHH, HIC, KJL and CCC designed this study. CWH, KLH and CCC, were responsible for diagnosis and clinical evaluation. SHH, HIC, SWH, KJL and TYH collected MRI and PET data. HIC and SWH analyzed MRI data. SHH, KJL and TYH analyzed PET data. CMC contributed to statistical analysis. MNL and CCC wrote the manuscript. CMC, CWH, HIC and KJL contributed to manuscript revisions. All authors read and approved the final manuscript.

Funding: This work was supported by the Chang Gung Memorial Hospital, Taiwan [grants number CORPG8N0041, CMRPG8J0524, CMRPG8J0843, CMRPG8K1533] and Ministry of Science and Technology, Taiwan [grants number 111-2314-B-182A-143, NSTC 112-2321-B-182A-004] to CCC.

Availability of data and materials

The data that support the findings of this study are not publicly available due to local regulations and to maintain the privacy of the research participants. However, the data are available from the corresponding author upon reasonable request.

Ethics approval and consent to participate

The study was approved by the Institutional Review Board of Chang Gung Memorial Hospital and all participants received written informed consent to participate.

Consent for publication

All authors have read and approved the final manuscript.

Competing interests

The authors declare that they have no competing interests.

Jack CR, Jr., Bennett DA, Blennow K, Carrillo MC, Dunn B, Haeberlein SB, et al. NIA-AA Research Framework: Toward a biological definition of Alzheimer's disease. Alzheimers Dement 2018, 14(4):535-562.
Jiang Z, Wang J, Qin Y, Liu S, Luo B, Bai F, et al. A nonhuman primate model with Alzheimer's disease-like pathology induced by hippocampal overexpression of human tau. Alzheimers Res Ther 2024, 16(1):22.
Yang Z, Sreenivasan K, Toledano Strom EN, Osse AML, Pasia LG, Cosme CG, et al. Clinical and biological relevance of glial fibrillary acidic protein in Alzheimer's disease. Alzheimers Res Ther 2023, 15(1):190.
Marizzoni M, Mirabelli P, Mombelli E, Coppola L, Festari C, Lopizzo N, et al. A peripheral signature of Alzheimer's disease featuring microbiota-gut-brain axis markers. Alzheimers Res Ther 2023, 15(1):101.
Mundada NS, Rojas JC, Vandevrede L, Thijssen EH, Iaccarino L, Okoye OC, Shankar R, et al. Head-to-head comparison between plasma p-tau217 and flortaucipir-PET in amyloid-positive patients with cognitive impairment. Alzheimers Res Ther 2023, 15(1):157.
Stern RA, Trujillo-Rodriguez D, Tripodis Y, Pulukuri SV, Alosco ML, Adler CH, et al. Amyloid PET across the cognitive spectrum in former professional and college American football players: findings from the DIAGNOSE CTE Research Project. Alzheimers Res Ther 2023, 15(1):166.
Ossenkoppele R, Schonhaut DR, Scholl M, Lockhart SN, Ayakta N, Baker SL, et al. Tau PET patterns mirror clinical and neuroanatomical variability in Alzheimer's disease. Brain 2016, 139(Pt 5):1551-1567.
Visser D, Verfaillie SCJ, Wolters EE, Coomans EM, Timmers T, Tuncel H, et al. Differential associations between neocortical tau pathology and blood flow with cognitive deficits in early-onset vs late-onset Alzheimer's disease. Eur J Nucl Med Mol Imaging 2022, 49(6):1951-1963.
Dubbelman MA, Mimmack KJ, Sprague EH, Amariglio RE, Vannini P, Marshall GA, et al. Regional cerebral tau predicts decline in everyday functioning across the Alzheimer's disease spectrum. Alzheimers Res Ther 2023, 15(1):120.
Kotari V, Southekal S, Navitsky M, Kennedy IA, Lu M, Morris A, et al. Early tau detection in flortaucipir images: validation in autopsy-confirmed data and implications for disease progression. Alzheimers Res Ther 2023, 15(1):41.
Lagarde J, Olivieri P, Tonietto M, Rodrigo S, Gervais P, Caille F, et al. Could tau-PET imaging contribute to a better understanding of the different patterns of clinical progression in Alzheimer's disease? A 2-year longitudinal study. Alzheimers Res Ther 2023, 15(1):91.
Bischof GN, Dodich A, Boccardi M, van Eimeren T, Festari C, Barthel H, et al. Clinical validity of second-generation tau PET tracers as biomarkers for Alzheimer's disease in the context of a structured 5-phase development framework. Eur J Nucl Med Mol Imaging 2021, 48(7):2110-2120.
Li L, Ji B, Zhao M, Bai L, Chen B. Nonfluent Variant Primary Progressive Aphasia on FDG, 11 C-PIB, and 18 F-APN-1607 PET Imaging. Clin Nucl Med 2023, 48(11):e539-e541.
Jia C, Wu M, Yen TC, Li Y, Cui R: Complementary Utility of Dopamine Transporter and Tau PET Imaging in the Diagnosis of Progressive Supranuclear Palsy: A Case Report. Clin Nucl Med 2022, 47(4):336-338.
Tagai K, Ono M, Kubota M, Kitamura S, Takahata K, Seki C, et al. High-Contrast In Vivo Imaging of Tau Pathologies in Alzheimer's and Non-Alzheimer's Disease Tauopathies. Neuron 2021, 109(1):42-58 e48.
Hsu JL, Lin KJ, Hsiao IT, Huang KL, Liu CH, Wu HC, et al. The Imaging Features and Clinical Associations of a Novel Tau PET Tracer-18F-APN1607 in Alzheimer Disease. Clin Nucl Med 2020, 45(10):747-756.
Lin HC, Lin KJ, Huang KL, Chen SH, Ho TY, Huang CC, et al. Visual reading for [(18)F]Florzolotau ([(18)F]APN-1607) tau PET imaging in clinical assessment of Alzheimer's disease. Front Neurosci 2023, 17:1148054.
Zhu XC, Tan L, Wang HF, Jiang T, Cao L, Wang C, et al. Rate of early onset Alzheimer's disease: a systematic review and meta-analysis. Ann Transl Med 2015, 3(3):38.
Joubert S, Gour N, Guedj E, Didic M, Gueriot C, Koric L, et al. Early-onset and late-onset Alzheimer's disease are associated with distinct patterns of memory impairment. Cortex 2016, 74:217-232.
Wattmo C, Wallin AK. Early- versus late-onset Alzheimer's disease in clinical practice: cognitive and global outcomes over 3 years. Alzheimers Res Ther 2017, 9(1):70.
Uddin MS, Hasana S, Hossain MF, Islam MS, Behl T, Perveen A, et al. Molecular Genetics of Early- and Late-Onset Alzheimer's Disease. Curr Gene Ther 2021, 21(1):43-52.
Ho GJ, Hansen LA, Alford MF, Foster K, Salmon DP, Galasko D, et al. Age at onset is associated with disease severity in Lewy body variant and Alzheimer's disease. Neuroreport 2002, 13(14):1825-1828.
Marshall GA, Fairbanks LA, Tekin S, Vinters HV, Cummings JL. Early-onset Alzheimer's disease is associated with greater pathologic burden. J Geriatr Psychiatry Neurol 2007, 20(1):29-33.
Liu Y, Tan Y, Zhang Z, Yi M, Zhu L, Peng W. The interaction between ageing and Alzheimer's disease: insights from the hallmarks of ageing. Transl Neurodegener 2024, 13(1):7.
Mudher A, Colin M, Dujardin S, Medina M, Dewachter I, Alavi Naini SM, et al. What is the evidence that tau pathology spreads through prion-like propagation? Acta Neuropathol Commun 2017, 5(1):99.
Weber CJ, Carrillo MC, Jagust W, Jack CR, Jr., Shaw LM, Trojanowski JQ, et al. The Worldwide Alzheimer's Disease Neuroimaging Initiative: ADNI-3 updates and global perspectives. Alzheimers Dement (N Y) 2021, 7(1):e12226.
Chang HI, Chang YT, Huang CW, Huang KL, Hsu JL, Hsu SW, et al. Structural Covariance Network as an Endophenotype in Alzheimer's Disease-Susceptible Single-Nucleotide Polymorphisms and the Correlations With Cognitive Outcomes. Front Aging Neurosci 2021, 13:721217.
Chang HI, Hsu SW, Kao ZK, Lee CC, Huang SH, Lin CH, et al. Impact of Amyloid Pathology in Mild Cognitive Impairment Subjects: The Longitudinal Cognition and Surface Morphometry Data. Int J Mol Sci 2022, 23(23).
Hughes CP, Berg L, Danziger WL, Coben LA, Martin RL. A new clinical scale for the staging of dementia. Br J Psychiatry 1982, 140:566-572.
Lin KN, Wang PN, Liu HC, Teng EL. Cognitive Abilities Screening Instrument, Chinese Version 2.0 (CASI C-2.0): administration and clinical application. Acta Neurol Taiwan 2012, 21(4):180-189.
Chang HI, Huang CW, Hsu SW, Huang SH, Lin KJ, Ho TY, et al. Gray matter reserve determines glymphatic system function in young-onset Alzheimer's disease: Evidenced by DTI-ALPS and compared with age-matched controls. Psychiatry Clin Neurosci 2023, 77(7):401-409.
Chang YT, Huang CW, Chen NC, Lin KJ, Huang SH, Chang WN, et al. Hippocampal Amyloid Burden with Downstream Fusiform Gyrus Atrophy Correlate with Face Matching Task Scores in Early Stage Alzheimer's Disease. Front Aging Neurosci 2016, 8:145.
Southekal S, Devous MD, Sr., Kennedy I, Navitsky M, Lu M, Joshi AD, et al. Flortaucipir F 18 Quantitation Using Parametric Estimation of Reference Signal Intensity. J Nucl Med 2018, 59(6):944-951.
Suchy-Dicey AM, Shibata D, Best LG, Verney SP, Longstreth WT, Jr., Lee ET, et al. Cranial Magnetic Resonance Imaging in Elderly American Indians: Design, Methods, and Implementation of the Cerebrovascular Disease and Its Consequences in American Indians Study. Neuroepidemiology 2016, 47(2):67-75.
Buckner RL, Head D, Parker J, Fotenos AF, Marcus D, Morris JC, et al. A unified approach for morphometric and functional data analysis in young, old, and demented adults using automated atlas-based head size normalization: reliability and validation against manual measurement of total intracranial volume. Neuroimage 2004, 23(2):724-738.
Braak H, Braak E. Neuropathological stageing of Alzheimer-related changes. Acta Neuropathol 1991, 82(4):239-259.
Cho H, Choi JY, Hwang MS, Kim YJ, Lee HM, Lee HS, et al. In vivo cortical spreading pattern of tau and amyloid in the Alzheimer disease spectrum. Ann Neurol 2016, 80(2):247-258.
Scholl M, Lockhart SN, Schonhaut DR, O'Neil JP, Janabi M, Ossenkoppele R, et al. PET Imaging of Tau Deposition in the Aging Human Brain. Neuron 2016, 89(5):971-982.
Zou KH, O'Malley AJ, Mauri L. Receiver-operating characteristic analysis for evaluating diagnostic tests and predictive models. Circulation 2007, 115(5):654-657.
Preacher KJ, Hayes AF. Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behav Res Methods 2008, 40(3):879-891.
Tian M, Zuo C, Civelek AC, Carrio I, Watanabe Y, Kang KW, et al. International Nuclear Medicine Consensus on the Clinical Use of Amyloid Positron Emission Tomography in Alzheimer's Disease. Phenomics 2023, 3(4):375-389.
Yamashita K, Miura M, Watanabe S, Ishiki K, Arimatsu Y, Kawahira J, et al. Fully automated and highly specific plasma beta-amyloid immunoassays predict beta-amyloid status defined by amyloid positron emission tomography with high accuracy. Alzheimers Res Ther 2022, 14(1):86.
Bullich S, Roe-Vellve N, Marquie M, Landau SM, Barthel H, Villemagne VL, et al. Early detection of amyloid load using (18)F-florbetaben PET. Alzheimers Res Ther 2021, 13(1):67.
Murugan NA, Nordberg A, Agren H. Different Positron Emission Tomography Tau Tracers Bind to Multiple Binding Sites on the Tau Fibril: Insight from Computational Modeling. ACS Chem Neurosci 2018, 9(7):1757-1767.
Mohammadi Z, Alizadeh H, Marton J, Cumming P. The Sensitivity of Tau Tracers for the Discrimination of Alzheimer's Disease Patients and Healthy Controls by PET. Biomolecules 2023, 13(2).
Ju Z, Li Z, Lu J, Jiao F, Lin H, Bao W, et al. In Vivo Tau Burden Is Associated with Abnormal Brain Functional Connectivity in Alzheimer's Disease: A (18)F-Florzolotau Study. Brain Sci 2022, 12(10).
Tagai K, Ikoma Y, Endo H, Debnath OB, Seki C, Matsuoka K, et al. An optimized reference tissue method for quantification of tau protein depositions in diverse neurodegenerative disorders by PET with (18)F-PM-PBB3 ((18)F-APN-1607). Neuroimage 2022, 264:119763.
Lee WJ, Cho H, Baek MS, Kim HK, Lee JH, Ryu YH, et al. Dynamic network model reveals distinct tau spreading patterns in early- and late-onset Alzheimer disease. Alzheimers Res Ther 2022, 14(1):121.
Frontzkowski L, Ewers M, Brendel M, Biel D, Ossenkoppele R, Hager P, et al. Earlier Alzheimer's disease onset is associated with tau pathology in brain hub regions and facilitated tau spreading. Nat Commun 2022, 13(1):4899.
Stern Y. Cognitive reserve in ageing and Alzheimer's disease. Lancet Neurol 2012, 11(11):1006-1012.
de Leon MJ, Pirraglia E, Osorio RS, Glodzik L, Saint-Louis L, Kim HJ, et al. The nonlinear relationship between cerebrospinal fluid Abeta42 and tau in preclinical Alzheimer's disease. PLoS One 2018, 13(2):e0191240.
Long JM, Holtzman DM. Alzheimer Disease: An Update on Pathobiology and Treatment Strategies. Cell 2019, 179(2):312-339.
Wu M, Zhang M, Yin X, Chen K, Hu Z, Zhou Q, et al. The role of pathological tau in synaptic dysfunction in Alzheimer's diseases. Transl Neurodegener 2021, 10(1):45.
Phillips JS, Nitchie FJt, Da Re F, Olm CA, Cook PA, McMillan CT, et al. Rates of longitudinal change in (18) F-flortaucipir PET vary by brain region, cognitive impairment, and age in atypical Alzheimer's disease. Alzheimers Dement 2022, 18(6):1235-1247.
Hammond TC, Xing X, Wang C, Ma D, Nho K, Crane PK, et al. beta-amyloid and tau drive early Alzheimer's disease decline while glucose hypometabolism drives late decline. Commun Biol 2020, 3(1):352.
Ridge PG, Ebbert MT, Kauwe JS. Genetics of Alzheimer's disease. Biomed Res Int 2013, 2013:254954.
Serrano-Pozo A, Qian J, Monsell SE, Blacker D, Gomez-Isla T, Betensky RA, et al. Mild to moderate Alzheimer dementia with insufficient neuropathological changes. Ann Neurol 2014, 75(4):597-601.
Aschenbrenner AJ, Gordon BA, Benzinger TLS, Morris JC, Hassenstab JJ. Influence of tau PET, amyloid PET, and hippocampal volume on cognition in Alzheimer disease. Neurology 2018, 91(9):e859-e866.
Lu J, Bao W, Li M, Li L, Zhang Z, Alberts I, et al. Associations of [(18)F]-APN-1607 Tau PET Binding in the Brain of Alzheimer's Disease Patients With Cognition and Glucose Metabolism. Front Neurosci 2020, 14:604.
Gao Y, Wang Y, Lei H, Xu Z, Li S, Yu H, et al. A novel transgenic mouse line with hippocampus-dominant and inducible expression of truncated human tau. Transl Neurodegener 2023, 12(1):51.
Morris GP, Clark IA, Vissel B. Questions concerning the role of amyloid-beta in the definition, aetiology and diagnosis of Alzheimer's disease. Acta Neuropathol 2018, 136(5):663-689

No competing interests reported.

9SupplementaryfiguretableAlzheimersResearchTherapy20240205.docx

Download PDF

Version 1

posted

You are reading this latest preprint version

Development and Clinical Validation of Global Tau Severity Score in Young- and Late- Onset Alzheimer's Disease Using Florzolotau (18F) PET

Status:

Version 1

Abstract

Background

Methods

Results

Conclusions

Figures

Introduction

Methods

Participants

Group stratification and diagnosis

Cognitive assessment

Data acquisition and image reconstruction

Pilot data: gTS score formula

Hippocampal volume and eTIV

Statistical analysis

Results

Pilot sample and gTS score formula

Demographic characteristics of validation dataset

The gTS score and Braak’s NFT signals of interest

Optimal gTS cutoff value for diagnosing YOAD or LOAD

The gTS score offers robust cognition prediction

Curve fitting differences in YOAD and LOAD

Mediation analysis

Discussion

Major findings

Optimal cutoff value to complement visual readouts and diagnose AD

Age of onset affect the influence of pathological tau severity to cognition

Distinct biological properties of tau on cognitive measurement in LOAD and YOAD

Biological properties of tau and the indirect mediation by hippocampus in salient features

Limitations

Conclusion

Abbreviations

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1