Early-phase 18F-FP-CIT and 18F-flutemetamol PET were significantly correlated

Little is known about whether early-phase PET images of 18F-FP-CIT match those of amyloid PET. Here, we compared early-phase 18F-FP-CIT and 18F-flutemetamol PET images in patients who underwent both within a 1-month interval. The SUVR on early-phase 18F-FP-CIT PET (median, 0.86) was significantly lower than that of 18F-flutemetamol PET (median, 0.91, p < 0.001) for total brain regions including all cerebral lobes and central structures. This significant difference persisted for each brain region except central structures (p = 0.232). The SUVR of total brain regions obtained from early 18F-FP-CIT PET showed a very strong correlation with that of 18F-flutemetamol PET (rho = 0.80, p < 0.001). Among the kinetic parameters, only R1 showed a statistically significant correlation between the two techniques for all brain regions (rho = 0.89, p < 0.001). R1 from 18F-FP-CIT (median, 0.77) was significantly lower in all areas of the brain compared to R1 from 18F-flutemetamol PET (median, 0.81, p < 0.001).18F-FP-CIT demonstrated lower uptake in cortical brain regions than 18F-flutemetamol on early-phase PET. However, both early-phase PETs demonstrated significant correlation of uptake.


Time-activity curves (TACs) from early-phase 18 F-FP-CIT and 18 F-flutemetamol PET scans. The
SUVR TACs from early-phase 18 F-FP-CIT and 18 F-flutemetamol PET fitted using a simplified reference tissue model (SRTM) are shown in Fig. 3. From 9 min onward, the 18 F-FP-CIT SUVR of the central structures was higher than that of the cerebral lobes (Fig. 3A). However, this pattern was not observed until 10 min in the SUVR TACs from 18 F-flutemetamol PET (Fig. 3B). Representative and typical SRTM fitting for SUVR TACs of a patient are shown in Fig. 3C,D. The individual SUVR TACs fitted using SRTM from 10 patients were provided in Supplementary Fig. 1. 18 F-flutemetamol PET scans. The delivery rate of 18 F-FP-CIT in total brain regions relative to the rate of delivery in the cerebellum (0.77 [0.68-0.83]), represented as R1, was significantly lower than that of 18 F-flutemetamol (0.81 [0.74-0.91], p < 0.001). This significant difference in R1 between the two PETs was consistent across all brain areas even when dividing by each region (all p < 0.05, Table 1). There was a very strong correlation in R1 between 18 F-FP-CIT and 18 F-flutemetamol PET in total brain regions (rho = 0.89, p < 0.001, Fig. 2B). A very strong correlation of R1 between the two PETs was seen Table 1. Early-phase parameters of 18 F-FP-CIT PET and 18 F-flutemetamol PET. *Interquartile range, † p-value from the Wilcoxon test for paired samples, ‡ regions including central structures and all cerebral lobes. SUVR (median, IQR*) R1 (median, IQR*) 18 F-FP-CIT PET 18  There were no significant differences or correlations in the efflux rate constant (k2) or binding potential (BP ND ) between the two PETs for any brain regions (Supplementary Table 1).

Discussion
We hypothesized at the beginning of this study that 18 F-FP-CIT and 18 F-flutemetamol activity in early-phase PET would be similar. However, comparing the two PETs showed that the cortical SUVR of 18 F-FP-CIT was significantly lower than that of 18 F-flutemetamol in the early phase, whereas there was no difference in SUVR in the central structures. To the best of our knowledge, no previous studies have compared early-phase PET using 18 F-FP-CIT and 18 F-flutemetamol. Therefore, it was difficult to find a precedent in the previous literature for the reasons underlying our results. The difference in SUVR between the two early-phase PET techniques is probably due to differences in their pharmacokinetic characteristics, and the apparently different shapes of TACs between the two PETs obtained in our study support this hypothesis. The SUVR of 18 F-FP-CIT in the central structures does not differ from that of 18 F-flutemetamol seems likely because of a steep increase in the activity of the central structures compared to other cortical regions on TACs. The central structures include the basal ganglia, the main target of 18 F-FP-CIT, so this is not surprising. Although cortical SUVRs varied between the two PETs, they showed a moderate or very strong correlation in all brain regions. Most previous studies that reported the usefulness of early-phase 18 F-FP-CIT or amyloid PET performed validation with 18 F-FDG PET 15,[19][20][21][22][23] or 15 O-water perfusion PET 24 , but our study did not. Patients included in our retrospective study did not undergo 18 F-FDG or perfusion PET imaging, so we cannot validate that early-phase PETs in our study reflect true brain perfusion. Further well-designed prospective studies including 18 F-FDG PET or perfusion PET are needed to validate the current study. However, based on our results, we suggest that if both 18 F-FP-CIT and 18 F-flutemetamol PETs are scheduled within a short period of time, early-phase imaging from only one technique would be sufficient because of the significant correlation in their uptake. In addition, it may be helpful if clinicians keep in mind that 18 F-FP-CIT may show lower uptake in cortical brain regions on early PET than 18 F-flutemetamol.
Another notable result in our study is that R1 obtained from dynamic data demonstrated a significant correlation between the two PETs. The R1 also showed a significantly lower value in 18 F-FP-CIT than in 18 F-flutemetamol PET, similar to SUVR, but there was a very strong correlation for most brain regions. The R1, which represents the delivery rate of radiopharmaceuticals to the regional brain, has recently been used as a proxy for measuring cerebral blood flow in early-phase PET with 18 F-flutemetamol 25 . In the central structures, SUVR showed no significant difference between the two PETs, but R1 was significantly different. It is difficult to clearly explain this discrepancy, but it is probably because the SUVR was obtained as the average value of the sum of the radiopharmaceutical activity over a 10 min duration, while the R1 value represents the delivery rate of radiopharmaceutical to the regional brain. On the other hand, k2 and BP ND obtained failed to show any significant correlation between the two early PET techniques. It could be that our early-phase scan time of 10 min was not sufficient to estimate k2 and BP ND . In fact, Heeman et al. 26 reported that a 60 min dual-time-window protocol of 0-30 and 90-110 min is needed to accurately estimate BP ND in 18 F-flutemetamol PET. Nevertheless, a strength of our study is that it demonstrated a significant correlation in the early phase of the two PETs with regard to the kinetic parameter R1 as well as SUVR. We would like to recommend pharmacokinetic modeling analysis in evaluating early phase PET www.nature.com/scientificreports/ images. Based on our results, it seems that it is necessary to evaluate early phase images to use pharmacokinetic modeling rather than simply to obtain SUVR. In our study, 18 F-flutemetamol was used as a radiopharmaceutical for amyloid PET. Previous studies that reported the usefulness of early-phase imaging with amyloid PET have used 11 C-Pittsburgh Compound B 19,22,24,27,28 , 18 F-florbetapir 20,23 , or 18 F-florbetaben 15,18,28,29 , and we could find only single previous report using 18 F-flutemetamol 26 . Since this study was carried out retrospectively, we could not select the radiopharmaceuticals used for amyloid PET. 18 F-flutemetamol was simply the main radiopharmaceutical used in our institution, so this study dealt with 18 F-flutemetamol. Thus, another strength of our study is that previous research reporting 18 F-flutemetamol early-phase PET is very rare.
There is not yet a clear consensus on the optimal acquisition time for early-phase brain PET for 18 F-FP-CIT and 18 F-flutemetamol. Jin et al. 17 conducted a study on the optimal time frame for 18 F-FP-CIT early-phase PET, and reported that the 10 min image was the most useful, whereas the quality of the image was too poor at the 5 min or 7 min time points. Heeman et al. 26 suggested the initial 30 min as the optimal time for early-phase 18 F-flutemetamol PET imaging. At our institution, obtaining an initial 10 min image from both PETs is a routine protocol. Since our method has not been proven, this was an obvious limitation of this study. Therefore, further research to determine the image acquisition time that best reflects the brain perfusion status of each radiopharmaceutical is needed.
There are several limitations to this study and they are as follows. First, the number of subjects included in this study is small. The statistical sample size was indeed satisfied, but we admit that 10 subjects was small. Due to the cost burden, it was not easy to find patients who needed both 18 F-FP-CIT PET and amyloid PET within the same month in our retrospective study. We look forward to future studies that will involve more subjects in order to validate our results. The second limitation was that we were unable to collect blood samples when acquiring dynamic images due to the retrospective research design. Therefore, we used SRTM, a kinetic model that can be used without blood sampling, which was also used in previous dynamic brain imaging studies 25,26,30 . www.nature.com/scientificreports/ In order to obtain results for other kinetic parameters that cannot be obtained from SRTM such as k1, future studies with blood sampling are warranted. The final limitation was that we could not enroll a homogeneous disease group. This study included patients with various diseases such as PD, PD with dementia (PDD), progressive supranuclear palsy (PSP), dementia with Lewy bodies (DLB), and AD. Although the disease groups varied, this should not present a major obstacle to comparing early uptake on PET performed at short intervals in the same patient, which was the goal of this study. However, studies in homogenous disease groups along with normal groups are needed to validate our results.
In conclusion, 18 F-FP-CIT exhibited a lower level of cortical uptake than 18 F-flutemetamol on early-phase PET, but uptake of both was significantly correlated.

Methods
Subjects. This study was conducted retrospectively. From September 2017 to September 2020, 15 patients were identified as having undergone both 18 F-FP-CIT PET and 18 F-flutemetamol PET from among the patient population at our single institution. All patients were clinically accompanied by cognitive impairment with parkinsonism symptoms, so both 18 F-FP-CIT PET and 18 F-flutemetamol PET were required. Of these, three patients who did not undergo early-phase PET imaging and two patients who did not have the magnetic resonance (MR) image data necessary for quantitative PET analysis were excluded. Finally, 10 patients (male/female = 6/4, median age 68 [IQR: 56-74] years, three patients with PD, three patients with PDD, two patients with PSP, one patient with DLB, and one patient with AD) were included. The interval between PETs for each patient was < 1 month (median 9 [IQR: 8-12] days). Also, MR images were acquired within 1 month of the PET images (median 6 [IQR: 5-11] days).
The clinical design of this retrospective study was approved by the Institutional Review Board of Ajou University (MED-MDB-20-511). The need for informed consent was waived.
Brain PET/CT acquisition. PET/computed tomography (CT) data were acquired on a Discovery ST scanner (GE Healthcare, Milwaukee, WI, USA). All patients were forbidden to take neurology-or psychiatric-related drugs for 24 h before PET examination. The radiopharmaceuticals were purchased from commercial companies [ 18 F-FP-CIT from DuChemBio (DuChemBi Co., Ltd., Seoul, South Korea) and 18 F-flutemetamol from GE Healthcare (Vizamyl, GE Healthcare, Seoul, South Korea)]. Their radiochemical purity was confirmed and specific activity at the end of synthesis was sufficiently satisfactory to be used for PET imaging before daily use. For early-phase imaging, brain CT (100 kV, 95 mA; section width = 3.75 mm) was obtained, then 10 min dynamic PET data [60 s per frame, three-dimensional (3D) mode] were acquired immediately after intravenous injection of each radiopharmaceutical . Routine delayed-image acquisition was started 90 min after injection of radiopharmaceuticals. The delayed PET data [10 min per frame of 1 bed duration for 18 F-FP-CIT and 20 min (4 × 5 min frames) for 18 F-flutemetamol, 3D mode] were obtained after brain CT (same parameters as early phase). All PET images were iteratively reconstructed (i.e., ordered subsets of expectation maximization with two iterations and 21 subsets, Gaussian filter (full width at half maximum = 2.14 mm), with a 128 × 128 matrix) from CT data for attenuation correction.
Quantitative analysis of early-phase PET images. All images were analyzed using Maximum Probability Atlas application in PMOD Neuro Tool (version 3.802, PMOD Technologies Ltd., Zurich, Switzerland). First, the averaged PET image was generated by averaging the frames from 0 to 10 min on the dynamic series. Then, the individual gray matter probability map was calculated by segmentation of each patient's T1-weighted MR image. The brain was split into left and right hemispheres and the cerebellum. MR images were spatially normalized to the Montreal Neurological Institute (MNI) T1 template. The segmented and normalized MR images were rigidly matched to the averaged PET image, and their alignments were visually checked by a specialist in nuclear medicine with 13 years of brain PET experience (YS An). The automated anatomic labeling (AAL)merged atlas 31 was transformed to MR space and cortical structures were intersected with the gray matter probability map (mask threshold of 0.3). The final VOIs applied to the matched PET series for calculating average regional uptake, represented as the standardized uptake value (SUV), were based on body weight. The VOIs of central structures, frontal, occipital, parietal and temporal lobe regions were selected. Averaged SUVs from each brain region were divided by averaged cerebellar SUV to obtain SUVR, and SUVR images were generated based on the method published by Peretti et al. 32 .
Also, the TAC of each region was obtained, and TACs were transferred to the kinetic modeling tool [PMOD Kinetic Modeling (PKIN)]. SRTM was developed with the cerebellum as a reference tissue. TACs fitted with SRTM and kinetic parameters including relative R1, k2, and BP ND were obtained using a coupled fit across the VOIs 33 . The detailed structures constituting each brain area are shown in Table 2, and the representative outline contours of VOIs for selected areas are shown in Fig. 4.
Statistical analysis. All statistical analyses were performed using MedCalc software (version 19.3.1; Med-Calc Software bvba, Ostend, Belgium). Power analysis was used to calculate the sample size required for this study using a significance (α) level of 5% and statistical power (1 − β) of 80%. A sample size of five for paired samples t test and nine for correlation coefficient test was required to obtain an appropriate confidence level; thus, our final enrolled number of subjects (n = 10) satisfied these requirements.
Data in our study did not follow a normal distribution as assessed by the Kolmogorov-Smirnov test. Therefore, all continuous variables are presented as the median and IQR, and appropriate nonparametric statistical methods were used to analyze the data. The Wilcoxon test for paired samples was used to determine whether a F-flutemetamol PET. The Spearman's coefficient for the ranked correlation test was used to assess the correlation of parameters between 18 F-FP-CIT and 18 F-flutemetamol PET. The magnitude of the correlation was interpreted as poor (|rho| < 0.3), fair (|rho| = 0.30-0.59), moderate (|rho| = 0.60-0.79), or very strong (|rho| ≥ 0.80) 34 . A p-value of less than 0.05 was considered statistically significant.

Ethics declarations
This retrospective study was conducted in accordance to the guidelines of the Declaration of Helsinki and approved by the Institutional Review Board of Ajou University (MED-MDB-20-511), through which informed consent was waived.

Data availability
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.