[18F]VUIIS1009B Features a Superior Imaging Performance to [18F]DPA-714 in TSPO Density Characterization for Neuroinammatory PET Imaging

Purpose: Translocator protein (TSPO), an outer mitochondrial membrane protein, is regarded as a key biomarker for neuroinammation in a variety of neurodegenerative diseases. In this study, we aim to evaluate two highly specic TSPO radiotracers [ 18 F]VUIIS1009A and [ 18 F]VUIIS1009B in a mild cerebral ischemic rat model, and to compare their in vivo performance to the well-established TSPO probe [ 18 F]DPA-714 for neuroinammation imaging. With multiple graphic analytical methods tested and macro parameters determined, we propose to nd a suitable and best quantication method to prole neuroinammation and measure TSPO density with the three TSPO radiotracers. Methods: Cerebral ischemia rat model was created and imaged using [ 18 F]VUIIS1009A, [ 18 F]VUIIS1009B and [ 18 F]DPA-714. Displacement studies using non-radioactive analogs were performed to evaluate the binding specicities of [ 18 F]VUIIS1009A and [ 18 F]VUIIS1009B individually. Imaging analysis using arterial plasma input functions (AIFs) was employed to generate Logan plots and parametric images of total distribution volume (V T ) for each radiotracer. Reference Logan model using contralateral brain as a reference region was introduced to generate parametric images for binding potential (BP ND ). Results: When compared to [ 18 F]DPA-714, [ 18 F]VUIIS1009B demonstrated higher binding potential (BP ND ) and distribution volume ratio (DVR). Parameter images of BP ND and V T also indicate [ 18 F]VUIIS1009B has a superior imaging prole with higher BP ND and DVR when compared with other two radiotracers in TSPO imaging. Correlation analysis between BP ND for [ 18 F]VUIIS1009B and [ 18 F]DPA-714 also indicates [ 18 F]VUIIS1009B is more sensitive than [ 18 F]DPA-714 in TSPO density measurement. Conclusions: This study demonstrates the superiority of [ 18 F]VUIIS1009B to [ 18 F]VUIIS1009A and [ 18 F]DPA-714 in the neuroinammation imaging. It also demonstrates that [ 18 F]VUIIS1009B PET imaging coupled with parameter mapping (V T and BP ND ) and graphic analysis using Logan analysis and reference Logan analysis holds great promise for neuroinammation characterization and TSPO density measurement. vivo displacement assays with unlabeled the displacement 66%) radiotracers, ratio 70%) [ displacement


Introduction
Neuroin ammation, as a response to local insult or the distally existing pathological events in central nervous system (CNS), often occurs in a variety of neurological disease states, including Alzheimer's diseases (AD), Parkinson's disease (PD), multiple sclerosis (MS), Huntington's disease (HD), and amyotrophic lateral sclerosis (ALS) [1][2][3][4] . Neuroin ammation often plays an important role in various neuropathologies. Neuroin ammation targeted diagnosis and therapy are often regarded to be crucial for disease progress monitoring and novel therapy evaluation [5] . Among those studies, neuroimaging provides a non-invasive tool to characterize and monitor in vivo neuroin ammation. As one of the neuroimaging modalities, positron emission tomography (PET) can be employed to visualize and quantify the target at nanomolar level, with both high sensitivity and speci city.
Translocator protein (TSPO, 18 kDa), one of the most popular PET imaging biomarkers for neuroin ammation, has been studied in neuroin ammation for over 20 years [6] . As an outer mitochondrial membrane protein, TSPO has an extremely low expression level in healthy brain [7] , but an overexpression when microglia are activated in neuroin ammation in response to brain injury [5] .
Although TSPO PET imaging has been performed in a variety of neuroin ammatory diseases, it still has some drawbacks, including relatively low neuroin ammatory uptake, low SNR and insensitivity to the small variance of TSPO expression when quanti ed with the semi-quantitative parameters (%ID/cc or SUV) [22] . In order to overcome these limitations, macro parameters like V T , BP ND and DVR were often introduced to TSPO imaging, with the aim to fully characterize radiotracer pharmacokinetics, TSPO expression level, as well as to increase SNR and to improve imaging visual effects [23] [22,[24][25][26] . In practice, radiotracers featuring higher binding speci city often demonstrate higher macro parameter values (like BP ND ), higher SNR and thus better visual effects in PET imaging as well as the higher sensitivity to TSPO density variance. Considering this, in this study, for the rst time we evaluated the performance of the highly speci c TSPO radiotracers [ 18 F]VUIIS1009A and [ 18 F]VUIIS1009B with directly comparison to the performance of [ 18 F]DPA-714 in a mild neuroin ammation model. As shown in Figure  1, VUIIS1009A (IC 50 : 14.4 pM) features a 750-fold higher in vitro TSPO binding a nity than DPA-714 (IC 50 : 10.9 nM) [27] . VUIIS1009B (IC 50 : 19.4 pM) features a 560-fold higher in vitro TSPO binding a nity than DPA-714. With in vivo PET dynamic scans, we evaluated their semi-quantitative parameters (%ID/cc or SUV), as well as macro parameters including binding potential (BP ND ), total distribution volumes (V T ) and distribution volume ratio (DVR). Speci c parametric images (BP ND , V T ) were also generated and compared to identify the suitable quantitative methods and parameters to pro le neuroin ammation in this study.

Materials And Methods
ll chemicals were purchased from commercial sources. Unlabeled DPA-714, VUIIS1009A/B and their precursor for radiosynthesis were synthesized in-house, and 18 F was produced using an IBA cyclotron (Belgium). E uent radioactivity was monitored using a NaI (Tl) scintillation detector system. All other synthesis reagents and solvents were purchased from Sigma-Aldrich (St. Louis, MO, USA) and used without further puri cation, unless noted otherwise.  [18,28] . In detail, precursors underwent nucleophilic uorination with uorine-18. Puri cation of 18 F-labeled radiotracers was performed with preparative HPLC. The retention time of all the radiotracers according to gamma detection was double checked to make sure it was in agreement with the UV retention time, which was determined using the corresponding non-radioactive analogs under the same condition for HPLC analysis. The radiochemical purity measured using HPLC was consistently greater than 99 %, with speci c activity consistently greater than 4203 Ci/mmol (156 TBq/mmol).
Animals. Animals were maintained and handled in accordance with the recommendations of the National Institute of Health in China. Animal studies were approved by the local University and Hospital Ethics Committee. All experiments conducted at the Shanghai Jiao Tong University School of Medicine were approved by the Animal Ethics Committee. Male Wistar rats (n = 9, 7 weeks old, 230 -250 g) were purchased from Vital River (Beijing, China) and housed under a 12-h/12-h dark/light cycle under optimal conditions. Ischemia Rat Models. Mild focal ischemia was induced by intraluminal occlusion of the middle cerebral artery for 30 min based on the intraluminal thread model [29] . After ischemic surgery, the rats were used for PET imaging, and metabolite analysis at 5 -7 days. Rats (n = 9) were a xed with arterial catheters prior to the PET/CT studies. Of the 9 animals, 6 rats were used repeatedly for PET imaging using the three individual radiotracers, 3 for repeatable displacement studies with [ 18 F]VUIIS1009A and [ 18 F]VUIIS1009B, with a total of 24 PET scans performed in this study. We chose 5-7 days for PET imaging with the aim to image the early stage of neuroin ammation instead of the advanced stage with maximal uptake. This early stage imaging is meaningful to early detection of neuroin ammation. Moreover, in the early stage of neuroin ammation, contralateral brain should have little or tiny expression of TSPO, which will bene t the following graphic analysis using it as the reference region.
In Vivo Dynamic PET Imaging Study. PET imaging was carried out using a small-animal PET scanner (Siemens Medical Solutions USA, Knoxville, TN, USA) according to the protocols reported previously [9] . Rats (n = 6) were rst anesthetised with iso urane (3.5% for induction, 1.5-2% for maintenance). Immediately after rats were intravenously injected via the tail vein with speci c radiotracers, a 45-min listmode emission scan was conducted immediately. The study was performed using [ 18  During the scans, blood samples were drawn according to the following schedule: 15 μl every 10 s for the rst 90 s and at 2, 5, 8, 12, 20, and 45 min. The time frame reconstruction for PET was as follows: 10s × 12 frames, 1 min × 3 frames, 5 min × 8 frames. In addition, the metabolite-corrected parent plasma input function (AIF) was measured according to our previous published protocol [17][18] .
For the displacement experiments, unlabeled TSPO compounds VUIIS1009A (10.0 mg/kg) or VUIIS1009B (10.0 mg/kg) were dissolved in 1.0 mL of saline containing 10% ethanol and 5% Tween-80, and injected 20 min after the PET scans were initiated. In this study, the same cohort of rats (n = 3) were imaged and evaluated for each radiotracer individually. Dynamic image reconstruction was achieved by ltered backprojection using Hanning's lter with a Nyquist cutoff frequency of 0.5 cycles/pixel.

PET Image Co-registration
In order to analyze the voxel values of the regions of interest (ROI) and calculate the parametric maps, PET images of the same rat with different radiotracers were co-registered. In detail, using Inveon Research Workplace 4.0 (Siemens Medical Solutions USA, Knoxville, TN, USA), corresponding CT images were rst co-registered to make the spatial alignment of brain regions. The transformation le for this coregistration was generated and then treated as an input to register the corresponding PET images. With the registered PET images, ROIs of ipsilateral and contralateral brain were drawn and values of voxels was analyzed.

Dynamic PET Data Analysis
The PET images were analyzed using PMOD version 3.4 image analysis software (PMOD Technologies, Zurich, Switzerland) and Inveon Research Workplace 4.0 (Siemens Medical Solutions USA, Knoxville, TN, USA). Regions of interest (ROIs) were manually de ned for each rat on the region of increased radiotracer binding in the ipsilateral hemisphere. The ROIs were manually drawn on ipsilateral and contralateral brain as shown in Figure 2 and Figure 3.
In this study, parametric V T images were generated using plasma input-based Logan analysis [30] , with the plasma contribution factor v p set to 0.05 for all the analysis. With the same animals, parametric BP ND images were also generated using Logan Reference Tissue model with contralateral brain TACs as an input [30] . For DVR LoganRef or BP ND determination and parametric image generation, k 2_REF was calculated using SRTM methods with contralateral brain TACs input as the reference tissue [31] . DVR Logan from the Table 1 was generated directly from the ratio of distribution volume of ipsilateral and contralateral brain generated via the plasma input-based Logan analysis.
Statistical Analyses. All quantitative data are expressed as the mean ± standard deviation (SD).
Pearson's correlation analysis was applied to investigate the relationship between different parameters (V T , BP ND , %ID/cc et al) both at the voxel level and regional level, using GraphPad Prism 5 software (GraphPad Software, La Jolla, CA, USA).

Results
In Vivo 45-min Dynamic Scan. Performance of [ 18 F]DPA-714, [ 18 F]VUIIS1009A and [ 18 F]VUIIS1009B was evaluated during the 45-min dynamic PET scan using a focal cerebral ischemic rat model, in which TSPO expression was up-regulated with activated microglia, whereas the blood-brain barrier (BBB) was intact or not seriously disrupted [32] . As expected, the ipsilateral side (ischemia side) showed a higher uptake of the radiotracer than the contralateral brain in the PET imaging for all of the three radiotracers, as shown in  , the uptake on the ipsilateral side was signi cantly higher than the uptake on the contralateral brain for both radiotracers before the injection of their nonradioactive analogs. However, after the displacement, radiotracer uptake of the ipsilateral brain dropped dramatically (Figure 3b and 3d), which accounts to more than 66% decrease of the tracer uptake when compared to the normal uptake in the same intervals as shown in the TACs for both radiotracers (as shown in Figure 2g, 2h, 3e and 3f). This is similar to the displacement ratio reported by Martin et al on [ 18 F]DPA-714 displacement study with the same ischemia animal model [33] . After displaced with their non-radioactive analogs, the uptake level of ipsilateral brain was almost comparable to the contralateral brain, both demonstrated by the imaging pro les and TACs in Figure 3. All these indicate the signi cantly high binding speci city for both radiotracers in ipsilateral brain.  Figure 2. In order to further evaluate their in vivo performances, macro parameters like V T , DVR and BP ND were determined using graphic analysis with AIFs or reference tissue TACs input. According to our previous studies, TSPO PET imaging can be analyzed using a two-tissue, four-parameter model [9,18,20] (model t as shown in Supplemental Figure 2). In this study, graphic analysis using AIFs can also be employed to determine the V T for both ipsilateral and contralateral brain region. As shown in Table 1, V T determined for ipsilateral region are higher than the corresponding values for contralateral brain for all the three radiotracers, indicating all these radiotracers tend to accumulate in ipsilateral region instead of the contralateral brain. Direct comparison of the ipsilateral V T from three radiotracers indicates [ 18 F]VUIIS1009B has a higher V T than the other two radiotracers as shown in Table 1. Moreover, for the contralateral brain, [ 18 F]VUIIS1009B also demonstrates a slightly higher V T than the other two radiotracers. The ratios of the V T values (noted as DVR Logan ) between ipsilateral brain and contralateral brain indicate a higher DVR Logan for Further study using contralateral brain as a reference region was also performed to determine DVR LoganRef and BP ND for all three radiotracers (as shown in Figure 4 and Table 1). Logan plots using contralateral brain as a reference region was plotted with BP ND and DVR LoganRef determined (Figure 4a, 4b and 4c), demonstrating a good t and linearity for all three radiotracers with r > 0.96. As shown in Table 1 In this study, we further compared the performance of the three radiotracers using parameter images (V T and BP ND ) generated ( Figure 5). In detail, voxel-wise V T images were generated via a Logan graphic method using AIFs. Voxel-wise BP ND images were generated using a Logan reference model with the contralateral brain input as the reference region. As

Correlation between %ID/cc, BP ND and V T at the Voxel Level
This study also determined and evaluated %ID/cc, BP ND and V T at the voxel level, with the aim to better elucidate the relationship among these parameters derived from the dynamic PET images. As shown in Figure 6, with %ID/cc, BP ND and V T determined for the voxels in the ipsilateral brain, we made different plots for the three probes and measured the correlation coe cient r and p value for each data set. As shown in Figure 6a and 6c, a strong correlation was elucidated for [ 18

Discussion
With a low expression in normal brain, TSPO is normally overexpressed in neuropathological conditions, such as stroke, brain trauma, AD, and PD. TSPO PET imaging is now becoming a useful tool in neuroin ammation evaluation, as well as in diagnosis and therapy evaluation for many neurological diseases. In the longitudinal studies of neurological diseases, TSPO density evaluation with different radiotracers and analytical approaches have been performed to increase the sensitivity of TSPO expression characterization, and thus to elucidate the relationship between the disease progression and TSPO density [34][35][36][37] . In these studies, semi-quantitative parameters like %ID/cc and SUV are widely evaluated in both clinical and preclinical studies [33,38] . Compared to the semi-quantitative parameters, macro parameters (like BP ND , DVR, V T ) can provide more detailed information on radiotracer pharmacokinetics and TSPO expression, which is more meaningful in longitudinal monitoring of neuroin ammatory diseases [39] . Moreover, TSPO radiotracers with higher binding a nities tend to have higher binding potential (BP ND ), which can be more sensitive to measure the variance of TSPO densities in the longitudinal analysis of disease progression. provides accurate informative information of the expression and distribution of TSPO activity after cerebral ischemia [33] . Similarly, Pulagam et al also demonstrated the feasibility to use [ 18 F]VUIIS1008 to monitor TSPO expression longitudinally in a rat model of cerebral ischemia, with the results suggesting 18 F-VUIIS1008 could become a valuable tool for the diagnosis and treatment evaluation of neuroin ammation following ischemic stroke [38] . Furthermore, Golla et al. used multiple methods to generate the quantitative images for [ 18 F]DPA-714 in a clinic study with healthy subjects and AD patients participated [40] . In this study, they concluded that both Logan analysis and spectral analysis are suitable plasma input-based methods to generate quantitatively accurate parametric V T images. Meanwhile, in reference tissue approaches, reference Logan analysis or SRTM2 can be used to generate parametric BP ND images [40] .
In our previous studies, we further modi ed [ 18 F]DPA-714 and discovered several novel TSPO imaging radiotracers, including [ 18 F]VUIIS1008 [16][17] , [ 18 F]VUIIS1018A [20][21] and [ 18 F]VUIIS1009A/B [18] . These novel radiotracers feature higher in vitro TSPO binding a nities and suitable lipophilicities for brain imaging, when compared to the parent radiotracer [ 18 F]DPA-714. Moreover, they also demonstrated great potential in TSPO PET imaging, including high SNR, higher binding potential, and enhanced visual effects [16][17][18] . In this study, we further evaluated the performance of [ 18 F]VUIIS1009A and [ 18 F]VUIIS1009B in a preclinical model of neuroin ammation. PET imaging shows both radiotracers can be employed to detect the neuroin ammation region with a high SNR, as well as a fast distribution and equilibration in both contralateral and ipsilateral regions. In order to con rm further the binding speci city of imaging [17][18] . We believe this behavior is due to the higher plasma protein binding a nity of  [18] . For [ 18 F]VUIIS1009A, the higher plasma protein binding a nity limits the partitioning of the radiotracer from blood to the tissues, and thus decreases its tissue uptake and image SNR, as well as the uptake of speci c binding compartment, which will then impact the following macro parameter determination.
In this study, by evaluating two highly speci c TSPO radiotracers [ 18

Declarations
Ethics approval and consent to participate.
All applicable international, national, and institutional guidelines for the care and use of animals were followed.  Figure 1 Structures of DPA-714, VUIIS1009A, VUIIS1009B and their corresponding IC50. Coronal and (f) transverse PET image of the same cerebral ischemia rat from dynamic scan with [18F]VUIIS1009B. Time activity curves (TACs) for the ipsilateral (red) and contralateral brain (blue) in the 45-min dynamic scan (n = 6) for [18F]DPA-714 (g), [18F]VUIIS1009A (h) and [18F]VUIIS1009A (f). %ID/cm3 = percentage injected dose per cubic centimeter. In TACs, data = mean ± SD. Ipsilateral ROI is marked by the red circle and contralateral ROI was marked by the yellow circle in the image.   Logan reference tissue method with TACs of the contralateral brain input as a reference region.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.
SI gs.pdf