18F-FIMP: a LAT1-specific PET probe for discrimination between tumor tissue and inflammation

Positron emission tomography (PET) imaging can assist in the early-phase diagnostic and therapeutic evaluation of tumors. Here, we report the radiosynthesis, small animal PET imaging, and biological evaluation of a L-type amino acid transporter 1 (LAT1)-specific PET probe, 18F-FIMP. This probe demonstrates increased tumor specificity, compared to existing tumor-specific PET probes (18F-FET, 11C-MET, and 18F-FDG). Evaluation of probes by in vivo PET imaging, 18F-FIMP showed intense accumulation in LAT1-positive tumor tissues, but not in inflamed lesions, whereas intense accumulation of 18F-FDG was observed in both tumor tissues and in inflamed lesions. Metabolite analysis showed that 18F-FIMP was stable in liver microsomes, and mice tissues (plasma, urine, liver, pancreas, and tumor). Investigation of the protein incorporation of 18F-FIMP showed that it was not incorporated into protein. Furthermore, the expected mean absorbed dose of 18F-FIMP in humans was comparable or slightly higher than that of 18F-FDG and indicated that 18F-FIMP may be a safe PET probe for use in humans. 18F-FIMP may provide improved specificity for tumor diagnosis, compared to 18F-FDG, 18F-FET, and 11C-MET. This probe may be suitable for PET imaging for glioblastoma and the early-phase monitoring of cancer therapy outcomes.


Results
As a result of screening our α-methyl amino acid chemical library using hLAT1 and hLAT2 overexpression cell lines, FIMP was found to have higher affinity for LAT1 than 2-aminobicyclo-(2,2,1)-heptane-2-carboxylic acid (BCH), a classical inhibitor of L-type amino acid transporters, which is also transported into cells as a substrate of LAT1. The half-maximal (50%) inhibitory concentration (IC50) value of FIMP was significantly lower than that of BCH (Mean ± SD, 88.5 ± 13.5 µM and 231.5 ± 10.0 µM, respectively).

Expression of human LAT1 and CD98 in cell lines and tumors. High expression levels of LAT1 and
CD98 were observed for T24 and LNZ308 tumor cells. Conversely, WI-38 normal human fetal lung fibroblast showed low to moderate expression of both proteins (Fig. 1a).
We also evaluated LAT1 and CD98 expression in tumor xenografts and normal muscle tissue. Moderate to high LAT1 and CD98 expressions were observed for T24 and LNZ308 xenografts (Fig. 1b). Conversely, LAT1 and CD98 expression was not detected for normal muscle tissue. Expression of sodium/potassium ATPase, plasma membrane markers, was comparable in all cells and tissues (Methods in Supplementary Information).
PET probe accumulation in tumor and inflamed tissue. PET probe accumulations were evaluated by small animal PET imaging using a mouse model which had both acute inflammation and a LAT1-positive tumor. After probe injection we observed time-dependent change of 18 F-FIMP accumulation by PET imaging in tumor, muscle and inflamed tissues. Time-activity curve of 18 F-FIMP accumulation in all tissues plateaued by 90 min after probe injection (Fig. 2). On the basis of this result, the biodistribution in all animals was evaluated at 90 min after probe injection. As a result of visual assessment, 18 F-FIMP showed high accumulation in the tumor region and low accumulation in the inflamed lesion. 18 F-FDG showed higher accumulation in tumor regions compared to 18 F-FIMP, 18 F-FET and 11 C-MET; however, comparatively high 18 F-FDG accumulation was also observed in inflamed lesions compared to 18 F-FIMP, 18 F-FET and 11 C-MET. 18 F-FET showed moderate accumulation in tumor regions and low accumulation in inflamed regions. 11 C-MET showed low accumulation in both tumor regions and inflamed regions (Fig. 3a). Furthermore, quantitative analyses of PET images found that accumulation levels of 18 F-FIMP in tumors (SUV 2.32 ± 0.09) were significantly higher than that of 18 F-FET (SUV 1.14 ± 0.20) and 11 C-MET (SUV 0.79 ± 0.14) (P < 0.01, respectively), comparable to that of 18 F-FDG (SUV 2.55 ± 0.59) without significant difference (P = 0.29). However, accumulation of 18 F-FIMP in inflamed lesions (SUV 0.96 ± 0.04) was comparable and significantly higher than that of 18 F-FET (SUV 1.14 ± 0.20) and 11 C-MET (SUV 0.79 ± 0.14) (P = 0.19 and P < 0.01, respectively), significantly lower than that of 18 F-FDG (SUV 1.73 ± 0.36) (P < 0.01) (Fig. 3b).
Tissue radioactivity was measured to validate the PET imaging data. Probe biodistribution data showed highest 18 F-FIMP uptake in the pancreas (33.99 ± 2.39%ID/g). Uptake of 18 F-FIMP was higher in tumor (7.78 ± 1.11%ID/g) than inflamed tissue, muscle tissue, and blood (3.89 ± 0.17, 2.36 ± 0.14, and   11 C-MET, and 18 F-FDG PET. LNZ308 cells were inoculated in right paws (arrow), and inflammation induced by injection of turpentine oil in left paws (arrow head). PET data were acquired 90 min after injection of 18 F-FIMP, 18 F-FET, and 11 C-MET, 18 F-FDG-PET data was acquired from 55 to 100 min after injection. Quantitative analysis of PET imaging data represented as (b) SUV and (c) tumor-to-muscle and inflamed lesion-to-normal muscle ratios (TMR and INR, respectively). Data are presented as mean ± SD (n = 4-9). *P < 0.05, **P < 0.01, compared with 18 F-FIMP groups.

18
F-FDG is the PET probe most commonly used for cancer diagnosis, staging, restaging, and assessment of therapy responses. However, the accumulation of this probe in inflamed lesions can lead to false positive diagnoses 1-3 . Therefore, efforts to further develop tumor-specific PET probes are important for increasing the effectiveness of cancer screening and treatment programs. In the present study, we demonstrated that the PET probe 18 F-FIMP The ratio of radioactivity in the unmetabolized fraction to that in total radioactivity was determined using a phosphoimaging plate at 90 min after injection into tumor-bearing mice. Data are presented as mean ± SD (n = 4). www.nature.com/scientificreports www.nature.com/scientificreports/ is highly specific for LAT1, with high accumulation in tumor tissue but not in inflamed lesions. T24 has higher expression of LAT1 than LNZ308 in tumor tissue (Fig. 1b). However, our final goal is to image brain tumors with 18 F-FIMP. So, we selected LNZ308 brain tumor cell line in this study, and used T24 as a positive control when examining LAT1 expression analyses. We first examined PET imaging in a subcutaneous tumor model implanted with LNZ308, and are currently considering PET imaging in a brain tumor model.
We confirmed that 18 F-FIMP is stable in aqueous 3.5% ascorbic acid. Starting with a radiochemical purity of 99.1%, this value decreased to 97.6% within 24 h after synthesis. For practical clinical and research applications, this high stability is advantageous as it allows for possible transportation of the synthesized 18 F-FIMP to hospitals and laboratories that are remote from the place of synthesis.
Inflammation is an inseparable by-product in the pathophysiology of cancer. Inflammation is not only a side-effect of cancer treatments, such as radio-and chemotherapy, but also contributes to the development and progression of cancer. At the earliest stages of neoplastic progression, inflammation can promote the progression of incipient neoplasia into invasive cancers 13 . Furthermore, inflammation can significantly hinder the efficacy of diagnostic tests. The accumulation of 18 F-FIMP in inflamed lesions was low in both animal models of inflammation (turpentine oil induced myositis model and collagen induced arthritis model); hence, suggesting that this probe may provide a more accurate approach to discriminate tumor tissues from inflamed tissues.
The efflux of radiolabeled amino acids and their metabolites from cells has been negatively correlated with their accumulation in tumors 14 . Most natural amino acid-derived PET probes, such as 11 C-MET, are incorporated into the protein fraction, resulting in an increase in nonspecific accumulation in non-tumor tissues 15 . From our results, 18 F-FIMP showed very high metabolic stability both under in vitro and in vivo conditions and is not incorporated into protein. These findings suggest that the high tumor accumulation value obtained using this tumor-specific PET probe maybe more reliable compared to that of other radiolabeled amino acids such as L-3-18 F-fluoro-α-methyl tyrosine (FAMT) 9 , O-(2-18 F-fluoroethyl)-L-tyrosine (FET) 16 , and anti-1-amino-3-1 8 F-flurocyclobutane-1-carboxylic acid (FACBC) 8 . In order to demonstrate whether 18 F-FIMP has superiority in tumor imaging, standard PET probes, such as 18 F-FDG and 11 C-MET, and 18 F-FIMP need to be directly compared in the same cancer inoculation models by PET imaging studies.
An accurate estimation of radiation exposure is indispensable for defining a safe clinical PET study protocol. According to the biodistribution data from our PET imaging studies, we were able to estimate the expected mean absorbed dose of 18 F-FIMP in humans. The effective doses of 18 F-FIMP were determined to be 25.1 ± 5.3 and 30.8 ± 6.7 μSv/MBq for males and females, respectively. These doses are comparable or slightly higher than that  17 but still indicate that 18 F-FIMP is a safe PET probe for use in humans.
However, additional studies are required to evaluate further applications of 18 F-FIMP in humans. For example, additional comparisons between 18 F-FIMP and other LAT1-specific PET imaging probes would provide a valuable assessment of its potential as a cancer diagnosis tool.
In this study, we developed a tumor imaging PET probe with a high affinity for LAT1. PET imaging studies revealed that 18 F-FIMP accumulated in LAT1-positive tumor tissue, but not in inflamed lesions. This markedly high discrimination between tumors and inflamed lesions is important for effective diagnosis and treatment. Hence, 18 F-FIMP may have advantages over existing PET imaging probes, such as 18 F-FDG and 11 C-MET.

Subcutaneous tumor xenograft and inflammation models.
All animal experimental protocols were approved by the Animal Care and Use Committee of RIKEN and performed in accordance with the National Institutes of Health Principles of Laboratory Animal Care (Approved No. MAH28-02). All applicable institutional and/or national guidelines for the care and use of animals were followed.
LAT1-positive human glioblastoma (LNZ308) cells were inoculated into the right forepaws of female BALB/ cAJcl-nu/nu nude mice (CLEA Japan, Inc., Tokyo, Japan) via subcutaneous injection of 5 × 10 6 cells in 100 μL phosphate buffered saline (PBS). Tumor growth was monitored twice weekly using a caliper. Acute-phase inflammation was induced by subcutaneous injection of 50 μL turpentine oil into the left forepaw of tumor-bearing mice 72 h before PET imaging 19 . We also used collagen-induced arthritis (CIA) model mice for evaluation of developed radiolabeled probes (see Supplementary Methods).
PET data acquisition. Mice were fasted for 14 h before 18 F-FDG administration. 18 F-FIMP and 11 C-MET-PET were administered to unfasted mice. All mice were anesthetized with 1.5% isoflurane and placed on the bed of a microPET Focus 220 scanner (Siemens Medical Solutions USA, Inc., Knoxville, TN). Radiolabeled probes were dissolved in saline (0.1 mL) and administered via a cannula inserted into the tail vein. Emission data were acquired for 90 min after administration using a 3-dimensional (3D) list-mode method for 18 F-FIMP and 11 C-MET, and for 45 min from 55 min after administration using a 3D list-mode method for 18 F-FDG. Data were reconstructed using 2-dimensional filtered back projection (FBP) for quantification and a 2-dimensional ordered subset expectation maximization (OS-EM) algorithm for region of interest (ROI) definition. For ROI definition and further analysis, summed images (0-90 or 55-100 min post injection) were reconstructed. ROIs were drawn on several areas of tumor, muscle, and inflamed tissues. Regional uptake of radioactivity in organs were decay-corrected based on injection times and expressed as the standardized uptake value (SUV), where SUV = tissue radioactivity concentration (MBq/cm 3 )/injected radioactivity (MBq) × body weight (g). Quantitative analysis of PET imaging data also represented as tumor-to-muscle and inflamed lesion-to-normal muscle ratios (TMR and INR, respectively). PET imaging was also performed with CIA mice using 18 F-FIMP, 11 C-MET and 18 F-FDG.
After PET imaging, mice were sacrificed and their organs quickly removed and washed with saline. Blood samples were obtained immediately before dissection by cardiac puncture. Excised organs and blood samples were weighed and their radioactivity determined using a Wallac Wizard 1480 scintillation counter (PerkinElmer, Waltham, MA). Results were expressed as %injected dose per gram of tissue, TMR, tumor-to-blood ratio (TBR), and INR. Dosimetry analysis. Mean absorbed doses of 18 F-FIMP (μSv/MBq) in humans were estimated on the basis of PET imaging data from mice (n = 4). The mean %ID/g values for mouse livers, kidneys, pancreases, urinary bladders, and remainder of the body were extrapolated to estimate expected uptake in organs for a 73 kg human adult male. The organ fractions of total body mass for mice (25 g), human males (73 kg), and human females (53 kg) required for this extrapolation were obtained from Hui et al. 20 for mice and ICRP Publication 60 17 for humans, respectively. Dosimetry estimations were calculated using the OLINDA/EXM version 1.1 software (Hermes Medical Solutions, Stockholm, Sweden) based on standard human male and female models 21 .
Statistical analysis. Data are presented as mean ± standard deviation (SD). All statistical analyses were performed using Microsoft Excel 2010 version 14.0 (Microsoft, Redmond, WA) using Student's t test. P-values less than 0.05 were considered significant.