18F-FPYBF-2, a new F-18-labelled amyloid imaging PET tracer: first experience in 61 volunteers and 55 patients with dementia

Objective Recently, we developed a benzofuran derivative for the imaging of β-amyloid plaques, 5-(5-(2-(2-(2-18F-fluoroethoxy)ethoxy)ethoxy)benzofuran-2-yl)-N-methylpyridin-2-amine (18F-FPYBF-2) (Ono et al., J Med Chem 54:2971–9, 2011). The aim of this study was to assess the feasibility of 18F-FPYBF-2 as an amyloid imaging PET tracer in a first clinical study with healthy volunteers and patients with various dementia and in comparative dual tracer study using 11C-Pittsburgh Compound B (11C-PiB). Methods 61 healthy volunteers (age: 53.7 ± 13.1 years old; 19 male and 42 female; age range 24–79) and 55 patients with suspected dementia [Alzheimer’s Disease (AD); early AD: n = 19 and moderate stage AD: n = 8, other dementia: n = 9, mild cognitive impairment (MCI): n = 16, cognitively normal: n = 3] for first clinical study underwent static head PET/CT scan using 18F−FPYBF-2 at 50–70 min after injection. 13 volunteers and 14 patients also underwent dynamic PET scan at 0–50 min at the same instant. 16 subjects (volunteers: n = 5, patients with dementia: n = 11) (age: 66.3 ± 14.2 years old; 10 males and 6 females) were evaluated for comparative study (50–70 min after injection) using 18F-FPYBF-2 and 11C-PiB on separate days, respectively. Quantitative analysis of mean cortical uptake was calculated using Mean Cortical Index of SUVR (standardized uptake value ratio) based on the established method for 11C-PiB analysis using cerebellar cortex as control. Results Studies with healthy volunteers showed that 18F-FPYBF-2 uptake was mainly observed in cerebral white matter and that average Mean Cortical Index at 50–70 min was low and stable (1.066 ± 0.069) basically independent from age or gender. In patients with AD, 18F-FPYBF-2 uptake was observed both in cerebral white and gray matter, and Mean Cortical Index was significantly higher (early AD: 1.288 ± 0.134, moderate AD: 1.342 ± 0.191) than those of volunteers and other dementia (1.018 ± 0.057). In comparative study, the results of 18F-FPYBF-2 PET/CT were comparable with those of 11C-PiB, and the Mean Cortical Index (18F-FPYBF-2: 1.173 ± 0.215; 11C-PiB: 1.435 ± 0.474) showed direct proportional relationship with each other (p < 0.0001). Conclusions Our first clinical study suggest that 18F-FPYBF-2 is a useful PET tracer for the evaluation of β-amyloid deposition and that quantitative analysis of Mean Cortical Index of SUVR is a reliable diagnostic tool for the diagnosis of AD. Electronic supplementary material The online version of this article (10.1007/s12149-018-1236-1) contains supplementary material, which is available to authorized users.


Introduction
Alzheimer's disease (AD) is the most common neurodegenerative disorder and the most common cause of dementia in the elderly, which affects 47 million patients worldwide with steadily increasing numbers [1]. The two characteristic neuropathological changes observed in AD are the deposition of extracellular amyloid senile plaques and the presence of intracellular neurofibrillary tangles (NFTs) [2,3]. The amyloid cascade hypothesis has been proposed and the deposition of amyloid beta (Aβ) protein is the first step of Alzheimer's pathology including NFTs [4]. Therefore, the deposition of Aβ protein has been the main target of in vivo diagnostic imaging tool of AD. Several imaging tracers, especially for positron emission tomography (PET), has been developed and reported to evaluate amyloid deposition, such as 11 C-Pittsburgh compound B (PiB) [5], 11 C-BF227 [6], 18 F-AZD4694 [7], 18 F-FACT [8], 18 F-BAY-949172 ( 18 F-florbetaben) [9], 18 F-AV-45 ( 18 F-florbetapir) [10], and 18 F-GE067 ( 18 F-Flutemetamol) [11]. PiB, the first amyloid imaging PET tracer, has been reported with successful results and used widely as a research tool [12]. However, the short half-life of labelled 11 C (20 min) limits the clinical utility of 11 C-PiB as a diagnostic tracer. Therefore, several 18 F-labelled beta amyloid tracers have been developed for commercial utility because 18 F with half-life 110 min has been recognized as commercially available radioactive tracer for clinical practice in terms of cost, supply and shipping.
Recently, we developed a benzofuran derivative for the imaging of Aβ protein, 5-(5-(2-(2-(2-18 F-fluoroethoxy)ethoxy)ethoxy)benzofuran-2-yl)-N-methylpyridin-2-amine ( 18 F-FPYBF-2) [13]. This new fluorinated benzofuran derivative, which is like 18 F-AZD4694 but has a fluoropolyethylene glycol side chain, is a promising PET probe for cerebral Aβ plaques imaging, and the specific labeling of Aβ plaques was observed in autoradiographic sections of autopsied AD brain. It should be noted that 18 F-FPYBF-2 has a stable chemical structure which does not photodegrade. However, there has been no report evaluating the utility of 18 F-FPYBF-2 as a PET tracer in in vivo human study.
The aim of this study was to assess the feasibility of 18 F-FPYBF-2 as an amyloid imaging PET tracer in a first clinical study with healthy volunteers and in comparative dual tracer study using 11 C-PiB, and to evaluate the clinical usefulness of 18 F-FPYBF-2 PET/CT in the diagnosis of AD.

Healthy volunteers
From March 2013 to July 2014, 61 healthy volunteers (male: 19, female: 42; mean age: 53.7 ± 13.1; age range: 24-79) ( Table 1) were included and underwent 18 F-FPYBF-2 PET or 18 F-FPYBF-2 PET/CT study as a   18 F-FPYBF-2 was performed using a modification of the methods described by Ono et al. [13] and on a hybrid synthesizer, cassette-type multipurpose automatic synthesizer module (JFE Engineering Corporation, Japan). 11 C-Pittsburg compound B ( 11 C-PiB) was also prepared in-house. The 11 C-CO 2 was produced with a cyclotron, CYPRIS HM18 [Sumitomo Heavy Industries (SHI), Ltd., Japan] by the 14 N(p, α) 11 C reaction on nitrogen gas (0.2% O 2 ). The radiosynthesis of 11 C-PiB was performed using a modification of the methods described by Verdurand et al. [17] and on a hybrid synthesizer, cassette-type multipurpose automatic synthesizer module (JFE Engineering Corporation, Japan).

PET data acquisition
In this first clinical volunteer study for newly developed 18 F-FPYBF-2, 61 cognitively healthy volunteers underwent 18 F-FPYBF-2 PET study (n = 28) or 18 F-FPYBF-2 PET/ CT (n = 33), respectively. PET scans were performed by a whole-body PET scanner, GE Advance (pixel size: 2 mm) (GE Healthcare, Waukesha WI, USA), while PET/CT scans were performed by a whole-body PET/CT scanner, Siemens True Point Biograph 16 (pixel size: 1.34 mm) (Siemens/ CTI, Erlangen, Germany). Static head PET image acquisition for 20-minutes was performed 50-70 min after the intravenous injection of 18 F-FPYBF-2 (200 ± 22 MBq). This 20-minute static scan was separately evaluated in two time zones (50-60 min and 60-70 min) for the evaluation of time interval difference in all cases. For the further evaluation of time-activity-curve (TAC) for the brain accumulation of 18 F-FPYBF-2, 50-minute dynamic PET/CT scan was also performed in 13 volunteers at 0-50 min at the same instant. In the present clinical patient study using 18 F-FPYBF-2 for 55 patients with suspected of dementia, 18 F-FPYBF-2 PET/CT were performed in all subjects. Static head PET image acquisition for 20-minutes was performed 50-70 min after the intravenous injection of 18 F-FPYBF-2 (204 ± 16 MBq) by a whole-body PET/CT scanner, Siemens True Point Biograph 16 (Siemens/CTI, Erlangen, Germany). This 20-minute static scan was also separately evaluated in two time-zones (50-60 min and 60-70 min) for the evaluation of time interval difference in all cases. For the further evaluation of time-activity-curve (TAC) for the brain accumulation of 18 F-FPYBF-2, 50-minute dynamic PET/CT scan was also performed in 14 patients with dementia at 0-50 min at the same instant.
In dual tracer study using 18 F-FPYBF-2 and 11 C-PiB for 5 volunteers and 11 patients with dementia or related disease, 18 F-FPYBF-2 PET/CT and 11 C-PiB PET/CT were performed in all subjects on separate days, independently. Static head PET image acquisition for 20-minutes was performed 50 min after the intravenous injection of 18 F-FPYBF-2 (213 ± 33 MBq) and 11 C-PiB (528 ± 57 MBq), respectively. The intervals between these two PET/CT studies were within 2 weeks for patients with dementia and half year for healthy volunteers. PET/CT scans were performed by a whole-body PET/CT scanner, Siemens True Point Biograph 16 (Siemens/CTI, Erlangen, Germany).
For the image data processing in both scanners (the PET scanner and the PET/CT scanner), further information was shown in Supplement.

C-PiB and 18 F-FPYBF-2 PET template construction
In-house PET template construction was performed for 11 C-PiB PET and 18 F-FPYBF-2 PET [18,19]. Further information was shown in Supplement.

Automated region of interest analysis
Since the cerebellar cortex can be used as a reference brain region lacking amyloid plaque [6,20], the Standardized Uptake Value Ratio (SUVR) of each region, indicating amyloid deposition, was calculated as follows; where SUV brain and SUV cerebellar cortex indicate SUV in each brain region and the cerebellar cortex, respectively. To obtain quantitative regional SUVR values of 18 F-FPYBF-2 PET and 11 C-PiB PET, we performed automated region of interest (ROI) analyses. The Automated Anatomical Labeling atlas (AAL) [21], which is publicly and widely available from the Internet, e.g. in open source software packages (MRIcro/MRIcron, http://www.mricr o.com/), were used as template-based predefined ROIs. The AAL atlas consists SUVR = SUV brain∕SUV cerebellar cortex, of 45 anatomical ROIs in each hemisphere and a cerebellar parcellation with 26 ROIs [22]. The AAL ROIs were finally masked with the gray matter defined by the MNI152 standard-space T1-weighted average structural template image available from the FSL software (http://www.fmrib .ox.ac.uk/fsl) and used as the predefined ROIs because the original AAL ROIs tend to be large and extend to the margin of the gray matter.
The reconstructed 18 F-FPYBF-2 PET and 11 C-PiB PET images were spatially normalized to a standard MNI space by the DCT-based approach [19] implemented in SPM8 with the in-house 18 F-FPYBF-2 PET and 11 C-PiB PET templates, respectively. The spatial normalization of amyloid PET images by the amyloid PET template was in accordance with the previous procedure for 11 C-PIB [23]. We also confirmed visually that inversely transformed AAL ROIs to an individual space corresponded to PET images in each subject. All AAL ROIs in the standard MNI space were inversely transformed to individual spaces by SPM8 using the inverse deformation field. Since these individual ROIs are automatically defined, the operator induced bias in defining ROIs manually can be avoided [24]. The cerebellar parcellation with 26 ROIs were combined and used as a reference region to create SUVR images. Mean SUVR values within 90 anatomical ROIs in both hemispheres were calculated by an in-house Matlab script.
Finally, as a representative value for cortical amyloid plaque deposition of each subject, the Mean Cortical Index was defined as a mean SUVR value within the frontal, posterior cingulate, precuneus, parietal and lateral temporal cortical regions [25].

Statistics
All values are expressed as mean ± SD. All the statistical analysis was performed using statistical software, JMP 8J version (SAS Institute, Cary NC, USA), in which p values < 0.05 were considered statistically significant. A comparison between each group was analyzed by the Wilcoxon or the Kruskal-Wallis Analysis for unpaired data. Correlation coefficient analysis between SUVR values of 18 F-FPYBF-2 and 11 C-PiB was performed by Pearson's analysis.

Volunteers
First clinical PET studies for volunteers were performed in 61 healthy volunteers. MMSE test showed high score in all the age range between 24 and 79 years old (Table 1). In this volunteer study, three subjects out of 64 volunteers were excluded. Further information was shown in Supplement.
In the evaluation of 50-min dynamic PET/CT scan, timeactivity-curve (TAC) for the cortical accumulation was calculated in 13 volunteers at the same instant and the SUV of cortex showed equilibrium phase with the similar level of that of cerebellar cortex at 50-70 min after injection of 18 F-FPYBF-2 (Fig. 2).

Patients
The 55 patients including 27 AD patients who were suspected of having dementia underwent 18 F-FPYBF-2 PET/ CT in clinical PET/CT study. MMSE test showed relatively higher score in MCI patients, while it showed lower score in AD and other dementia patients ( Table 2). No adverse events were reported during at least 1-year follow-up period.
Clinical studies with 27 AD patients showed that 18 F-FPYBF-2 uptake was observed in cerebral gray matter as well as cerebral white matter (Fig. 1). In the evaluation of 50-minute dynamic PET/CT scan, TAC for the cortical accumulation of AD patients showed higher retention and reached higher equilibrium level than that of cerebellar cortex (Fig. 2). Average Mean Cortical Index at 50-70 min after injection were high in AD patients with healthy volunteer (upper) and Alzheimer disease (moderate stage) (bottom). Regional TACs were shown for cerebral cortex (including frontotemporal lobe), white matter, cerebellar cortex, thalamus and pons. Subjects who showed similar TACs to mean TACs of each group were selected. Radioactivity was shown as Standardized Uptake Value (SUV) in this figure. In AD patients, TAC of cortex showed higher retention than that in healthy volunteers, and reached a plateau similar to the level of thalamus and pons 1 3 moderate (1.342 ± 0.191) and early stage (1.288 ± 0.134), which were significantly higher than those of healthy volunteers (49 years old or younger and 50 years old or older) (p < 0.01, p < 0.001) (Fig. 3). There was no significant difference between the Mean Cortical Index of early stage AD and moderate stage AD patients.

Subjects of dual tracer study
Sixteen subjects underwent both 18 F-FPYBF-2 PET/CT study and 11 C-PiB PET/CT study on separate days, respectively (Table 3). Final diagnosis of these 16 subjects were as follows; healthy volunteer: n = 5, MCI: n = 3, early stage AD: n = 4, moderate stage AD: n = 1, other dementia (amyloid angiopathy, FTD and unknown): n = 3. In this table, the diagnostic classification (high or WNL) was performed based on the quantitative analysis of Mean Cortical Index. Average Mean Cortical Index at 50-70 min after injection were higher at 11 C-PiB PET/CT study (1.435 ± 0.474), while its standard deviation was wider. On the other hand, average SUVR at 18 F-FPYBF-2 PET/CT study was lower and stable with its smaller standard deviation (1.155 ± 0.219). Mean Cortical Index of SUVR at both studies showed an excellent linear correlation between each other (correlation coefficient, r 2 = 0.87, p < 0.0001) (Fig. 4 left). The results in Fig. 4 left indicate that 18F-FPYBF-2 has lower lesion-to-normal contrast than 11C-PiB both in SUVR and in images. Regression line in Fig. 4 left showed that Mean Cortical Index of 1.2 in 18 F-FPYBF-2 corresponded to Mean Cortical Index of 1.5 in 11 C-PiB. Uptake pattern of 18 F-FPYBF-2 and 11 C-PiB in each brain region was also similar with each other. Using this threshold value for 18 F-FPYBF-2, Mean Cortical Index of 1.2, the average SUVR for amyloid positive cases in AD (n = 20) and healthy volunteers excluding amyloid positive cases in volunteer study (n = 57) were 1.372 ± 0.110 and 1.054 ± 0.055 (mean ± SD), respectively. Figure 4 right showed brain images of three cases for example: a case of normal volunteer, a case of moderate stage AD, and a case of early stage AD with a history of traumatic subarachnoid hemorrhage. In the third case, amyloid deposition in left occipital lobe was prominent both at 18 F-FPYBF-2 PET/CT study and 11 C-PiB PET/CT study.

Discussion
Our first clinical study clearly indicated that 18 F-FPYBF-2 is a safe and stable amyloid PET tracer with longer half-life with F-18 and is comparable to 11 C-PiB in the detectability of amyloid deposition with high linear correlation, and that 18 F-FPYBF-2 PET/CT is a useful and reliable diagnostic tool for the evaluation of AD. Although 18 F-FPYBF-2 is a "late" amyloid PET tracer after the appearance of several tracers in clinical practice with comparable diagnostic ability, we would like to show the potential of 18 F-FPYBF-2 as diagnostic abilities as an amyloid imaging tracer and expand the utilization of this tracer further in various fields of research and clinical practice in the following sentences. The diagnostic abilities of 18 F-FPYBF-2 as an amyloid imaging tracer are satisfactory and comparable to the other amyloid PET tracers (Figs. 1, 3). Figure 3 showed that differential diagnosis between AD patients and healthy volunteers was achieved using the qualitative analysis of Mean Cortical Index of SUVR, and that the threshold of Mean Cortical Index was about 1.2. Figure 4 left also showed that Mean Cortical Index of 1.2 in 18 F-FPYBF-2 closely corresponded to Mean Cortical Index of 1.5 in 11 C-PiB. It is known that Mean Cortical Index of 1.5 has been used as a threshold between AD patients and healthy volunteers in 11 C-PiB PET/CT study [26,27]. Although the threshold value is different, we believe that diagnostic performance of 18 F-FPYBF-2 PET/CT using quantitative analysis of Mean Cortical Index is enough in the differential diagnosis of AD from healthy volunteers. We have to admit that the threshold value of Mean Cortical Index 1.2 could not clearly separate AD and healthy volunteers. In the present study, there were several AD patients with low amyloid deposition and several persons with high amyloid deposition in healthy volunteer group. However, we believe that this is reasonable in clinical study. As we mentioned above, our clinical diagnosis of AD was not confirmed with pathology or others. In addition, it is known that about 10-30% of healthy aged persons showed high amyloid deposition [28]. Therefore, we may say that our data of amyloid positive rate are comparable with the previously published data with other amyloid tracers.
While typical cases shown in Fig. 1 can be clearly diagnosed visually, we were under impression that visual diagnosis would not be easy in 18 F-FPYBF-2 PET/CT. As shown in Fig. 4 and Table 3, Mean Cortical Index of AD patients observed at 18 F-FPYBF-2 PET/CT is relatively lower than . In the third case, amyloid deposition in left occipital lobe was prominent compared to right lobe both at 11 C-PiB and 18 F-FPYBF-2. Please note that window level of SUVR was set 0-3.0 for 18 F-FPYBF-2 because of its narrow dynamic range, while it was set 0-4.0 for 11 C-PiB that at 11 C-PiB PET/CT, which means that 18 F-FPYBF-2 has a narrower dynamic range and lower lesion-to-normal contrast than 11 C-PiB both in SUVR and in images. It may be said that the impact of 18 F-FPYBF-2 as a PET imaging tool of AD is not so attractive in a visual sense compared to that of 11 C-PiB.
Several amyloid PET tracers have been developed and comparative study between 11 C-PiB and each PET tracer was performed so far [29][30][31][32][33]. While some of the tracers also showed the narrower dynamic range than that of 11 C-PiB, most of these reports revealed that diagnostic abilities of these amyloid PET tracers were similar and identical to that of 11 C-PiB. In these PET tracers, its own specific method for visual and quantitative diagnosis is proposed in each tracer, including the indication of color or black and white tones for image display, the indication of analyzed area of brain, etc. We have to admit that we could not establish the most appropriate specific visual diagnostic method for 18 F-FPYBF-2 PET/CT, so far. For the establishment of visual diagnosis, further evaluation of appropriate color scaling with comprehensive interpretation or new diagnostic method with regional area analysis or others would be needed.
It is particularly worth noting that evaluation of brain amyloid deposition in healthy volunteers was performed in broad spectrum of age range from 24 to 79 in age. Our data showed that average Mean Cortical Index of healthy volunteers (20-39, 40-49, 50-59, 60-69, 70-79 years old) were almost similar, except for the difference between the group 20-39 years old and the group 70-79 years old (p < 0.05) (Supplement Fig. 1). A slight upward extension of the distribution was due to appearance of high SUVR subjects in the older age range. This is reasonable because it is known that, as the age increases, some normal subjects present amyloid deposition while others remain amyloid negative. Our data of 8 healthy young volunteers in the age of 20-39 can be used as a control group for the evaluation of premature senility, which is often observed in patients with Down syndrome. It is known that adults with Down syndrome even at its younger age are at a very high risk of developing early onset AD due to trisomy of chromosome 21 [34]. We are planning to have further research using 18 F-FPYBF-2 PET/ CT as a diagnostic tool for the evaluation of early onset AD in Down syndrome in near future.
The limitation of the present study should be addressed. First of all, three-dimensional magnetic resonance (MR) images were not available in the study. Because the predefined AAL ROIs were masked with the gray matter of averaged standard T1-weighted structural template image in all subjects, the SUVR values in the study were influenced by inter-subject variability of gray matter volumes and the partial volume effect. Furthermore, the spatial normalization in the study was based on amyloid PET images by the amyloid PET template. Although we visually confirmed that inversely transformed AAL ROIs to an individual space corresponded to PET images in each subject, the accuracy of spatial normalization might be lower compared with MRbased normalization. The more sophisticated methods of spatial normalization only using PET images would improve both the accuracy of spatial normalization and quantification [35][36][37]. Second, the clinical diagnosis of AD patients and other related disease was performed before PET study by board certified physicians in a comprehensive diagnosis using Japanese clinical diagnostic guideline. Therefore, there was no confirmation of the diagnosis by pathological examination or autopsy, or biomarkers in cerebrospinal fluids, such as Aβ40, Aβ42 and phosphorylated Tau (pTau). Third, in the present study, 16 patients of Mild Cognitive Impairment (MCI) showed inconclusive results of Mean Cortical Index. The results of Mean Cortical Index of SUVR in MCI patients showed relatively high and could not be clearly distinguished with those of AD patients. However, we believe that this result was reasonable. MCI was just a diagnosis at the time before PET study. It is possible that within these MCI patients there must be several patients included who will develop AD in future. Further follow-up study would be needed to clarify the outcome of the patients with high amyloid deposition.
In conclusions, our first clinical study showed that 18 F-FPYBF-2 is a safe and stable amyloid PET tracer with longer half-life with F-18 and its diagnostic ability is comparable to 11 C-PiB. In addition, it can be said that 18 F-FPYBF-2 PET/CT is a useful and reliable diagnostic tool for the evaluation of AD by the quantitative analysis using Mean Cortical Index of SUVR, which could clearly distinguish Alzheimer disease patients by threshold of 1.2.