Specificity of translocator protein-targeted positron emission tomography in inflammatory joint disease

Objective Expression of the translocator protein (TSPO) on inflammatory cells has facilitated imaging of synovitis with TSPO-targeted positron emission tomography (PET). We aimed to quantitatively assess the specificity of the second-generation TSPO PET radioligand, [11C]PBR28, and to generate simplified PET protocols in patients with inflammatory joint disease (IJD) in this pilot study. Methods Three IJD patients (two rheumatoid arthritis and one osteoarthritis) with knee involvement underwent dynamic [11C]PBR28-PET scans before and after administration of 90 mg of oral emapunil (XBD-173), a TSPO ligand the same day. Radial arterial blood sampling was performed throughout the scan, and total radioactivity and radioactive metabolites were obtained. A semi-automated method was used to generate regions of interest. Standardized uptake value (SUV) and SUV ratio corrected for activity in bone and blood between 50 and 70 min (SUVr50–70 bone, SUVr50–70 blood, respectively) and PET volume of distribution (VT) of the radioligand were calculated. Results A mean [11C]PBR28 radioactivity of 378 (range 362–389) MBq was administered. A significant decrease (p < 0.05) in VT, SUVr50–70 bone and SUVr50–70 blood observed after oral emapunil confirmed the TSPO specificity of [11C]PBR28. A decrease in SUV was not observed in the post-block scan. Conclusion [11C]PBR28 is TSPO-specific radioligand in IJD patients. Simplified PET protocols with static PET acquisition can be used in the management and evaluation of novel therapeutics that target TSPO overexpressing cells.


Introduction
Synovial envelope of the articular joint is a critical provider of synovial fluid components and articular cartilage nutrients. Synovial inflammation (synovitis) together with progressive degeneration of articular cartilage is key pathological features in a variety of inflammatory joint diseases (IJDs), as diverse as rheumatoid (RA), psoriatic, juvenile and idiopathic arthritis, lupus and gout and a significant contributor of articular cartilage degeneration in osteoarthritis (OA) [1]. Synovial inflammatory infiltrate is composed of aggressive macrophage-and fibroblast-like mesenchymal cells, macrophage-like cells, fibroblast-like synoviocytes and other inflammatory cells [2,3].
Detection and treatment of subclinical or early inflammatory arthritis is likely to prevent disease progression, permanent joint damage and associated comorbidity. However, subclinical or early disease is often difficult to detect, leading to a delay in diagnosis [4]. Furthermore, the assessment of response to treatment of inflammatory disease is often based on composite disease activity scores, which can be highly subjective and difficult to reproduce consistently. Therefore, an unmet clinical need is to functionally evaluate the target tissue and assess changes in the infiltrate not only for quantitative assessment but also to evaluate targeted therapies.
Positron emission tomography (PET) studies with the 18 KDa translocator protein (TSPO) radioligands have Open Access *Correspondence: a.saleem@hull.ac.uk demonstrated the high expression of TSPO (formerly known as the peripheral benzodiazepine receptor; PBR) on activated macrophages at sites of inflammation, and indeed, up-regulation of TSPO has been noted in macrophages of inflamed synovium of animal models of inflammatory arthritis [5]. Clinical TSPO-PET studies in healthy volunteers and subjects with rheumatoid and psoriatic arthritis have reported higher TSPO radioligand uptake in IJD subjects compared to healthy volunteers [6,7]. However, to date there have been no clinical studies to confirm that the elevated PET signal can be attributed to specific binding to TSPO in IJD subjects. In this manuscript, we report on a cohort of subjects with IJD who underwent dual [ 11 C]PBR28-PET scans prior to and after a heterologous oral TSPO blocking agent, emapunil (XBD-173), to assess the specificity of TSPO binding in the knee joints of the subjects scanned.

Regulatory approval and subjects
The study was approved by the West Midlands-Black  (Table 1). Participants provided written informed consent, with eligibility determined by medical history, physical examination, coagulation screen and blood genotyping for rs6971single-nucleotide polymorphism [8]. Subjects with homozygous G allele (high-affinity binders; HABs) and heterozygous G/A allele at position 439 (Ala147Thr) (medium-affinity binders; MABs) were only included in the study as they have higher binding to [ 11 C]PBR28 [8].

PET scanning
[ 11 C]PBR28 PET scanning was performed at Invicro, London, on a Siemens PET/CT system Biograph 6 TruePoint with TrueV or a Hi-Rez Biograph 6 (Siemens Healthcare, Erlangen, Germany). Prior to the scan, all patients had a radial arterial cannulation for sampling of blood during the scan and a venous cannula for administration of radioactivity. [ 11 C]PBR28 was manufactured as described previously [9].
Initially, a low-dose CT scan of the both knee joints was performed for localization and attenuation correction followed by a 90-min dynamic PET scan after intravenous administration of [ 11 C]PBR28. Continuous arterial blood sampling was performed for the initial 15 min; additionally, discrete blood samples were performed for the full duration of the scan to obtain total blood and plasma radioactivity and the fraction of parent radiolabelled compound in the blood. A second [ 11 C]PBR28 PET-CT scan was performed the same day after oral administration of the TSPO ligand. Emapunil (XBD-173) was given per orally about 2 h before the second scan.

Data analysis
PET images were reconstructed with corrections applied for attenuation, randoms and scatter. All PET-CT images acquired were converted to the NIfTI (Neuroimaging Informatics Technology Initiative) format, and CT images were re-sliced to match the PET image resolution. Synovial regions of interest (ROIs) for each knee were generated using a global threshold method on CT scan, based on the Hounsfield Unit values for bone to exclude bone and adjoining tendon and muscle. For each knee, a single ROI for each knee was generated to include all of the inter-osseous area exclusive of muscle and tendons (Fig. 1). Finally, a manual review of the ROIs was performed by one operator and reviewed by another. Each knee was analyzed separately due to the variation in the severity of joint disease with osteoarthritis and rheumatoid arthritis, as done previously [6].  Tissue time-activity curves (TAC) were generated [10] as previously described for each joint synovia (Fig. 2). Radial arterial blood data obtained were used to generate plasma [ 11 C]PBR28 TAC for the full duration of the scan (input function) and was modeled with the tissue data (output) (Fig. 3) to generate volume of distribution (V T ) values using a two-tissue compartment model (2TC) as described previously [11,12].

Statistical analysis
Paired t tests were used to compare V T , SUV or SUVr values within subjects. p values less than 0.05 were considered statistically significant.

PET analysis
The composite PET image for the full duration of the scan showed radioactivity uptake in the synovia of knee joints, with minimal uptake in the adjoining bone. Review of the TACs (Fig. 2) demonstrated a plateauing in [ 11 C]PBR28 SUV between 50-and 70-min post-radioligand injection; hence, SUV and SUVr were calculated between these time points for all subjects. Variability of radioactivity uptake between the two knees post-block was minimal compared to pre-block uptake. V T was obtained in all subjects by fitting the 2TC model, apart from one subject (subject 3, Baseline), where the model did not fit the data well.
There was no change in mean (SD) SUV 50-70 after oral XBD-173 (0.65 (0.21) versus 0.69 (0.13); p = NS). In contrast, PET parameters that account for peripheral distribution of the radioligand such as V T , SUVr 50-70 blood and SUVr 50-70 bone showed a significant decrease in radioligand uptake in the synovium with mean (SD) respective pre-and post-block ligand uptake for V T being  Table 2. Figure 4 illustrates a decrease in [ 11 C]PBR28 uptake in subject 2 after normalization Fig. 3 The percentage of parent [ 11 C]PBR28 contributing to the total radioactivity (a) and the plasma over blood ratio (b) in arterial blood that were used to generate the plasma [ 11 C]PBR28 input function for the full duration of the scan (c) for subject 1 at baseline and post-block to blood activity (SUVr 50-70 blood) after dosing with emapunil.

Discussion
We conducted this study to evaluate if the increased uptake of [ 11 C]PBR28 observed in the joints of patients with IJD was specific to TSPO binding and not a consequence of increased radioactivity in joints due to other associated pathological processes. The decrease in the [ 11 C]PBR28 quantitative uptake parameter, V T , after administration of a heterologous TSPO blocking agent confirmed that the increase in [ 11 C]PBR28 uptake was specific to TSPO in the tissue cellular infiltrate. However, performing dynamic PET imaging with radial arterial sampling may not always be appropriate as the small joints of the wrist are commonly involved in RA and need to be imaged. Further, performing scans of shorter duration without arterial blood sampling would increase subject tolerability. Since the semiquantitative parameter SUV, which does not account for changes in peripheral metabolism of the radioligand [13], did not exhibit a decrease in uptake after blocking of TSPO, we investigated if we could use a pseudo-reference agent as in the brain where the cerebellum has been used [14]. We used both blood and cancellous bone, both of which contain TSPO [15,16] as pseudo-reference region as an estimate specific [ 11 C]PBR28 tissue uptake. We found that like V T , a decrease in SUVr 50-70 blood and SUVr 50-70 bone was observed after TSPO blocking with XBD-173 was noted with a decrease, indicating that both SUVr 50-70 blood and SUVr 50-70 bone could be used to quantify [ 11 C]PBR28 uptake. We also confirmed that a static [ 11 C]PBR28 scan between 50 and 70 min, correcting for nonspecific binding using bone SUV (SUVr 50-70 bone) as a pseudo-reference region, allows quantification of TSPO expression with [ 11 C]PBR28. This is likely to help facilitate PET studies that provide better subject comfort to evaluate disease and TSPO-targeting therapy and also allow assessment of disease in the small joints of the hand in subjects with IJD.

Conclusion
In conclusion, we observed that [ 11 C]PBR28 behaves as a TSPO-specific radioligand in the knee joint synovia of patients with IJD. PET uptake of [ 11 C]PBR28 can also be quantified using simplified acquisition of static PET scans, likely to be suitable in the evaluation of novel therapeutics that target TSPO overexpressing cells in IJD.