The imaging features and clinical associations of a novel tau PET tracer - 18F-APN1607(18F-PM-PBB3) in Alzheimer’s disease

Background: In vivo tau positron emission tomography (PET) imaging could help clarify the spatial distribution of tau deposition in Alzheimer’s disease (AD) and aid in the differential diagnosis of tauopathies. To date, there have been no in vivo 18 F-APN1607 tau PET studies in patients with AD. Methods: We applied tau tracer in twelve normal controls (NCs) and ten patients in the mild to moderate stage of probable AD. Detailed clinical information, cognitive measurements and disease severity were documented. Regional standardized uptake value ratios (SUVRs) from 18 F-AV-45 (florbetapir), 18 F-APN1607 PET images and regional gray matter (GM) atrophic ratios were calculated for further analysis. Results: Quantitative analyses showed significantly elevated SUVRs in the frontal, temporal, parietal, occipital lobes, anterior and posterior cingulate gyri, precuneus, and parahippocampal region (all ps < 0.01) with medium to large effect sizes (0.44 - 0.75). The SUVRs from 18 F-APN1607 PET imaging showed significant correlations with the ADAS-cog scores (all ps < 0.01) and strong correlation coefficients (R squared ranged from 0.54 to 0.68), even adjusted for age and gender effects. Finally, the SUVRs from 18 F-APN1607 PET imaging of the parahippocampal region showed rapid saturation as the ADAS-cog scores increased, and the SUVRs of the posterior cingulate gyrus and the temporal, frontal, parietal and occipital regions slowly increased. The combined SUVRs from 18 F-AV-45 PET, 18 F-APN1607 PET and regional GM atrophic ratio showed that uptake associated with the amyloid burden rapidly increased and reach a plateau, whereas uptake associated with tau depositions increased slowly and finally followed by regional GM atrophic ratios in most regions as the ADAS-cog scores increased. However, different regions exhibited various combinations of these patterns. Conclusions: Our findings suggest that the 18 F-APN1607 tau tracer showed a clear All Neurological examinations performed on all participants. Each participant completed the following components: screening evaluation, brain magnetic resonance imaging (MRI), 18 F-AV-45 (florbetapir) PET and 18 F-APN1607 PET. The screening procedures included vital signs, electrocardiography (ECG), physical examinations and laboratory tests. In addition, 18 F-AV-45 PET imaging results were used as part of the inclusion criteria to confirm the presence and absence of amyloid deposition in patients with probable AD and in NCs. All participants completed a series of clinical assessments and clinical safety studies to ensure that they were medically stable after participating in this study. A final follow-up phone call for adverse event assessment was made within 7 days following 18 F-APN1607 PET imaging. There were no adverse or clinically detectable pharmacologic effects in any participant. No significant changes in vital signs or the results of laboratory studies or ECG were observed, either. shows the significant correlation between the regional SUVRs of the posterior cingulate gyrus and the ADAS-cog scores. Age, gender and

reconstructed using a 3-D ordered-subset expectation maximization algorithm (4 iterations, 24 subsets; Gaussian filter: 2 mm; zoom: 3) with computed tomography-based attenuation correction and with the scatter and random correction procedures provided by the manufacturer. The reconstructed images had a matrix size of 400 × 400 × 148 and a voxel size of 0.68 × 0.68 × 1.5 mm 3 . 18 F-AV-45 PET scans were performed one month before the 18 F-APN1607 PET scans. The 18 F-AV-45 radiosynthesis and PET data acquisition were performed according to our previous protocols [32,33]. Briefly, all participants underwent 18 F-AV-45 PET scans on a Biograph mCT PET/CT System. PET images were acquired after intravenous injection of 374 ± 21 MBq of 18 F-AV-45. A 10 min scan was acquired starting at 50 mins after the tracer injection. PET images were reconstructed using the same method described above, and the images were then reconstructed with a matrix size of 400 × 400 × 148 and a voxel size of 0.68 × 0.68 × 1.5 mm 3 .

Image analysis
All imaging data were transformed into the Neuroimaging Informatics Technology Initiative (NIFTI) format using the MRIcron tool (http://www.mccauslandcenter.sc.edu/mricro/mricron/) for further processing. For each participant, PET images (both 18 F-AV-45 and 18 F-APN1607 images) and T1-weighted images in native space were analyzed. We coregistrated each PET image to individual MRI using the SPM12 toolbox (http://www.fil.ion.ucl.ac.uk/spm/software/spm12/) [34]. This procedure ensured the 18 F-AV-45 PET and 18 F-APN1607 images in alignment with the native MR images. The Muller-Gartner method was used for partial volume correction [35].
Then, the high-resolution MR images in native space were normalized to the Montreal Neurological Institute (MNI) standard space with the DARTEL toolbox in SPM12 [36]. This transform matrix was applied to PET images. The averaged intensity across the whole cerebellum was used as the reference for the 18 F-AV-45 PET images, and the cerebellum gray matter considered having no amyloid and tau pathology in AD was used as the reference region for the 18 F-APN1607 PET images [37]. Eighteen regions of interest (ROIs), including the bilateral frontal, parietal, temporal, occipital lobes, anterior and posterior cingulate gyrus, precuneus, hippocampus, and the parahippocampus were selected based on the Harvard-Oxford cortical structural atlas, and the average values from both sides were used for further study [38]. Finally, the regional standardized uptake value ratios (SUVRs) from both 18 F-AV-45 PET images and 18 F-APN1607 PET images were calculated using the mean intensity in the target ROIs divided by the averaged intensity of the corresponding reference regions [39]. Analysis of gray matter (GM) volume was performed on T1-weighted MRI using the Computational Anatomy Toolbox (CAT), and comparison of two groups were performed to search for significant atrophic regions (http://www.neuro.uni-jena.de/cat/index.html#About) [40]. To study the regional GM atrophy in patients with AD and NCs, we firstly calculated the modulate GM volumes in the target ROIs and divided by the individual total intra-cranial volume (ICV) as regional GM ratios. In NCs, the mean value from each regional GM ratio was treated as the benchmark to explore the relative atrophy in patients with AD and NCs. Finally, the regional atrophic ratio in each ROI was calculated using the formula: regional atrophic ratio = (mean regional GM ratio -individual regional GM ratio)/mean regional GM ratio.

Statistical analysis
All statistical analyses were performed using SPSS (version 21.0, Chicago, IL). Continuous variables were expressed as the mean ± standard deviation (SD). Nonparametric Mann-Whitney U tests and chi-squared/Fisher's exact tests were performed to compare age and gender distributions between AD patients and NCs. In the MRI study, a significant level of GM atrophy between the two groups was defined as an uncorrected p-value < 0.01 with a corresponding t-value > 2.54 and a cluster size > 100 voxels. For PET analysis, the effect sizes of Mann-Whitney U test of regional SUVRs in both 18 F-AV-45 PET images and 18 F-APN1607 PET images were measured by η 2 (range 0-1) as described in the previous literature [41,42]. Pairwise correlation using Spearman's rho was used to study the associations of regional SUVRs between the 18 F-AV-45 PET images and the 18 F-APN1607 PET images. To study the associations between cognition and the regional SUVRs derived from the 18 F-AV-45 PET images and 18 F-APN1607 PET images, we performed regression analyses. To study the sequential changes of regional SUVRs in the 18 F-AV-45 PET images, 18 F-APN1607 PET images and regional atrophic ratio from MRI images, we applied a nonlinear curve fitting model using the software GraphPad Prism, version 5.0 (GraphPad Inc., San Diego, CA). Statistical significance was defined as a p-value < 0.01.

Demography
The demographic information of the twelve participants with probable AD and ten NCs was described in Table 1. The mean age of patients with probable AD was older than that of NCs (mean age of probable AD patients: NCs = 75.2 ± 10.0: 56.0 ± 11.8, p < 0.01). The mean interval from disease onset to scanning time in patients with probable AD was 6.1 ± 2.4 years. No significant group differences in gender, ApoE4 genotype and total ICV differences were found (p = 0.39, p = 0.63 and p = 0.19, respectively). Significantly lower MMSE and higher ADAD-cog and CDR-SB scores were found in patients with probable AD than in NCs (all ps < 0.01). Nonparametric Mann-Whitney U tests revealed significantly lower GM ratios in the parietal, temporal, occipital, posterior cingulate gyrus, precuneus, hippocampus and parahippocampus of probable AD patients than those of NCs (all ps< 0.01, Supplementary table 1). Figure 1A shows four representive cases of 18 F-APN1607 PET images in NC, and patients with probable AD with mild or moderate stages. Upon visual inspection of 18 F-APN1607

Visual description of 18 F-APN1607 PET images in probable AD patients and NCs
PET images in NCs, there were no prominent hyperintensities in the cortical regions ( Figure 1B). The cerebral white matter, midbrain and basal ganglia also showed no significant uptake. In five of twelve NCs, the mean choroid plexus revealed approximately 2.5-5 times higher SUVRs than the reference regions. In patients with probable AD, the regions showing the most significantly increased uptake were the precuneus, the parietal, temporal, and frontal regions, and the parahippocampal region. The medial occipital region and the insular cortex showed weakly increased tracer uptake compared with the reference regions. The choroid plexus showed increased tracer uptake in seven of ten patients with probable AD. As for GM, patients with probable AD showed significant GM atrophy in the bilateral medial temporal, precuneus and parietal regions, a topographical distribution similar to that of tau deposition from averaging 18 F-APN1607 PET images of all patients with probable AD( Figure 1C-D).

Regional differences of SUVRs in 18 F-APN1607 and 18 F-AV-45 PET images
Nonparametric Mann-Whitney U tests were performed to study the regional differences in AD showed significantly higher SUVRs in the frontal, parietal, temporal, and occipital lobes, the anterior and posterior cingulate gyri, and the precuneus region (all ps < 0.01).
There were no significant group differences in the hippocampal and parahippocampal regions (p = 0.08 and 0.81, respectively). The effect size values from 18 F-AV-45 PET images were smaller than those from 18 F-APN1607 PET images in most regions. Table 3 shows the results of pairwise correlations of regional SUVRs derived from the 18 F-AV-45 PET images and the 18 F-APN1607 PET images. The values of Spearman's rho (rankcorrelation coefficient) showed significant associations in the frontal, temporal, parietal, and occipital lobes, the anterior and posterior cingulate gyri, and the precuneus region.
Interestingly, the SUVRs of the parahippocampus from the 18 F-APN1607 PET images had significant associations with those of all the above regions (all ps < 0.01), but the values from the 18 F-AV-45 PET images did not. The SUVRs from the hippocampal region showed no significant associations with any of the regions. These results demonstrated that similar trends of the tau and amyloid depositions between the parahippocampus and the rest of the studied brain regions but the hippocampus failed to show the same pattern.

Correlation studies between regional SUVRs and clinical parameters
To explore the correlations between regional SUVRs and clinical scores in 18 F-APN1607 PET images, we performed regression analyses in patients with probable AD and in NCs.
The SUVRs of the frontal, parietal, temporal, and occipital lobes, the anterior and posterior cingulate gyri, the precuneus, and the parahippocampal regions showed significant correlations with the ADAS-cog scores (all ps < 0.01). The values of R squared ranged from 0.54 to 0.68. The hippocampus did not show a significant association with the ADAScog scores (p = 0.53). Figure 2 shows the significant correlation between the regional SUVRs of the posterior cingulate gyrus and the ADAS-cog scores. Age, gender and ApoE4 gene were used as covariates in the regression model, and there were no significant associations with regional SUVRs (p = 0.23, p = 0.67 and p = 0.85, respectively).The CDR-SB showed significant associations with the above regions, and the values of R squared ranged from 0.52 to 0.61 (all ps < 0.01), except for the hippocampus regions (p = 0.77).
In 18 F-AV-45 PET images, regional SUVRs of the frontal, parietal, temporal, and occipital lobes, the anterior and posterior cingulate gyri, and the precuneus region showed significant associations with the ADAS-cog scores (all ps < 0.01). The regional SUVRs from the parietal, temporal, and occipital lobes, the posterior cingulate gyrus, and the precuneus region showed significant associations with CDR-SB (all p < 0.01).

Relations between regional SUVRs in 18 F-APN1607 and 18 F-AV-45 PET images and cognitive status
To further explore the relationship between regional SUVRs in 18 F-APN1607 PET images and the ADAS-cog scores, we used the sigmoidal four-parameter logistic curve fitting model( Figure 3A). The SUVRs in the parahippocampal region rapidly increased values as the ADAS-cog scores increased and then reached a plateau. This was followed by increased SUVRs of the posterior cingulate gyrus and the temporal, frontal, parietal and occipital regions, whose values sequentially increased as the ADAS-cog scores increased ( Figure 3B). Quantitative analysis indicated that the ADAS-cog scores at the inflection points of the sigmoidal curves from the above regions showed the lowest value in the parahippocampus (20.3), followed by the precuneus (38.6), temporal lobe (39.9), posterior cingulate gyrus (42.5), frontal lobe (42.5), parietal lobe (45.7), anterior cingulated gyrus (51.5) and occipital lobe (56). Figure 4 shows the combination of SUVRs from the 18 F-AV-45 and 18 F-APN1607 PET images and regional atrophic ratios in the different ROIs. In most regions, the SUVRs from the 18 F-AV-45 PET images rapidly increased as the ADAS-cog scores increased, except the parahippocampus region( Figure 4A), which didn't show increased uptake as the ADAS-cog scores increased. The SUVRs in most ROIs from the 18 F-APN1607 PET images showed gradual increases and reached plateaus as the ADAS-cog scores increased, except the occipital region. The regional atrophic ratios from T1weighted MRI showed flatter curves of increase, compared with the curves from the 18 F-APN1607 PET images as the ADAS-cog scores increased, except the parahippocampus region.

Discussion
In the current work, we applied the most recently developed tau tracer, 18

Significant associations between regional SUVRs in 18 F-APN1607 PET and clinical scores
In previous AD studies, cognitive decline and tau accumulation showed a close shown on 18 F-AV-45 PET images also showed significant associations with the ADAS-cog scores and CDR-SB scores, which may be related to the small sample size in this study.

The sequential changes in regional SUVRs from 18 F-APN1607 PET imaging
The pathological study showed that the spread of tau deposits started from the entorhinal cortex (Braak stages I/II), moving to the inferolateral temporal cortex and parts of the medial parietal lobe (stages III/IV), and eventually spreading throughout the association cortex (V/VI)[20, 58]. Our results using in vivo 18 F-APN1607 PET images demonstrated a similar topographical pattern. At least three patterns of tau deposition could be found ( Figure 3A). The first pattern was in the parahippocampal region; tau deposition rapidly increased and then reached a plateau (rapid saturation) as the ADAS-cog scores increased. The second pattern was in the posterior cingulate gyrus and the temporal, frontal and parietal regions, undergoing a slow progressive increase in tau deposits and then reaching to plateaus. The final pattern was in the occipital region, which showed a gradual increase of tau deposition without a plateau (Figure 3). The ADAS-cog scores at the inflection points of the sigmoidal curves showed the lowest value in the parahippocampus, followed by the precuneus, temporal lobe, posterior cingulate gyrus, frontal lobe, parietal lobe, anterior cingulated gyrus and occipital lobe. These findings were in agreement with the previous neuropathological evidence of neurofibrillary changes from transentorhinal stages to limbic stages and finally to neocortical stages [20].

The evolution of amyloid, tau and atrophic changes in different regions
In most of the ROIs, the evolution of SUVRs from 18 F-AV-45 and 18 F-APN1607 PET images and regional atrophic ratios showed that the amyloid burden usually rapidly increased to a plateau as the ADAS-cog scores increased, especially in the low ranges of ADAS-cog scores. The patterns of increasing tau deposition were regionally dependent. Finally, the regional atrophic ratios from MRI showed progressively increased values without plateaus ( Figure 4).
When we combined the SUVRs from 18 F-AV-45 and 18 F-APN1607 PET images and regional atrophic ratio information in the same ROIs to explore the sequential changes, we found that the amyloid burden usually manifested earlier than tau deposition and tau deposition usually started earlier then regional atrophies in most regions (Figure 4). In the parahippocampal region, tau deposition and regional atrophic ratio rapidly increased in the low ADAS-cog scores range but the amyloid burden didn't show a significant increase( Figure 4A). On the other hand, the occipital region showed a progressive increase of tau deposition and regional atrophy without a plateau phase( Figure 4B). These findings may indicate that cerebral amyloid deposition reached a saturation state more rapidly than tau deposition and neurodegeneration in most areas, but this sequential change also had the regional variability. Currently, our findings from the cross-sectional data could demonstrate the importance of amyloid-tau-neurodegeneration (ATN) sequential changes in regional base level, which were compatible with the widely hypothesized model of AD and ATN classification system in the AD research framework[59, 60].

Limitations
Several limitations of the current work need to be addressed. First, the tau tracer 18 F-APN1607 is a relatively new tracer, thus the pathological results are not yet available in our study. Up to now, only postmortem brain tissue had been studied with this tracer, and there have been no clinicopathological correlation studies using this tracer yet [21].
Furthermore, the six isoforms of tau in the brain include 3R and 4R tau, whose misfolding is responsible for various neurodegenerative diseases, such as progressive supranuclear palsy, corticobasal syndrome and frontotemporal dementia [61]. Whether the tau tracer 18 F-APN1607 can differentiate among all isoforms is an open question that needs further investigation. In addition, direct application of MAO-B inhibitors in patients undergoing 18 F-APN1607 PET imaging has not been performed, and it could be difficult to eliminate these concerns about the first-generation tau tracers [15]. Second, our study had a small sample size, a significant age difference between AD patients and NCs, and no participants with amnestic mild cognitive impairment. We acknowledge the demographic differences between groups, and we used age and gender as covariates to study the correlations of regional SUVRs from 18 F-APN1607 PET imaging with ADAS-cog and CDR-SB scores. Increasing the sample size and adding amnestic patients will help us explore the features of this tau tracer. Third, we used the ADAS-cog scores as the severity index for curve fitting with the regional SUVRs from 18 F-AV-45 , 18 F-APN1607 PET images and regional atrophy ratios. We acknowledge that any biomarker changes to be incorporated into the hypothetical model of AD should come from longitudinal studies rather than crosssectional observations, and our findings must be interpreted conservatively. Future studies should focus on longitudinal changes in 18 F-APN1607 PET imaging with the aid of other biomarkers, which will may provide further evidence for the AD hypothetical model.

Conclusions
This is the first in vivo study of the PET tracer 18 F-APN1607 in patients with mild to moderate AD. Our findings suggest that 18 F-APN1607 PET imaging has a clear background and no off-target binding in the basal ganglia or the thalamus. The regional SUVRs of the AD-associated regions were significantly correlated with cognitive deficits and disease severity. Finally, combined tau imaging with information on amyloid deposition and neurodegeneration may further our understanding of dynamic biomarker changes in the regional base level during the progression of AD.      Significant correlation between the ADAS-cog scores and regional SUVRs determined from 18F-APN1607 PET images of the posterior cingulate gyrus.  The evolution of increased SUVRs from 18F-AV-45 PET, 18F-APN1607 PET imaging and regional atrophic ratios in (A) the parahippocampus, the precuneus, the temporal and the posterior cingulated gyrus region and (B) the frontal, the parietal, the anterior cingulate gyrus and the occipital region. The SUVRs of 18F-AV-45 PET and 18F-APN1607 PET and regional atrophic ratios were fit for the ADAS-cog scores. In most regions, the amyloid burden showed rapid saturation as the ADAS-cog scores increased, while uptake associated with tau depositions were slowly increased. Finally, the regional atrophic ratios were gradually increased. ACG: anterior cingulated gyrus; PCG:posterior cingulated gyrus.