First-in-human PET study of [18F]FSW-100 for imaging histone deacetylase 6 in the brain: safety, biodistribution, radiation dosimetry, and brain kinetic analyses Authors: Tetsuro Tago, Muneyuki Sakata, Yuto Kamitaka, Masashi Kameyama, Kenji Ishii, Jun Toyohara Published: 2026 ABSTRACT BackgroundHistone deacetylase 6 (HDAC6) is a lysine deacetylase that modulates protein function and turnover, and is critically implicated in the pathogenesis of neurodegenerative disorders. [18F]FSW-100 is a promising radioligand designed to enable visualization of HDAC6 in the brain. In this first-in-human PET study, we evaluated the clinical applicability of [18F]FSW-100 by comprehensively assessing its safety profile, biodistribution, radiation dosimetry, and brain kinetics.MethodsIn cohort A, whole-body PET was performed in three healthy men to evaluate radiation dosimetry, with serial imaging conducted from 1 min to 2 h after injection followed by additional scans at 4 h and 6 h. Absorbed doses in individual organs and the effective dose were estimated. In cohort B, five healthy men underwent 120-min dynamic brain PET with arterial blood sampling for radioactivity measurement and metabolite analysis. To determine optimal kinetic modeling, one- and two-tissue compartment models (1-TCM and 2-TCM), along with Logan graphical analysis (LGA), were applied to estimate distribution volume (VT). Safety was monitored throughout both cohorts.ResultsThe scan protocol was well tolerated by all subjects. In cohort A, effective dose was estimated at 18.9 ± 1.7 µSv/MBq, with the highest absorbed dose observed in the gallbladder wall (379 ± 136 µGy/MBq). Accordingly, as a precaution to minimize gallbladder dose, cohort B received a lower administered radioactivity. In brain PET analysis, 2-TCM better described radioligand kinetics than 1-TCM. VT was highest in cerebellar gray matter (42.6 ± 9.7 mL/cm3) and lowest in white matter (12.2 ± 3.6 mL/cm3). VT values from 2-TCM strongly correlated with those from LGA. The mean VT coefficient of variation across regions was 26 ± 3%, decreasing to 19 ± 3% when the input function was derived from averaged unchanged fractions, and further to 9 ± 2% when VT values were normalized to white matter.Conclusions[18F]FSW-100 PET demonstrated an acceptable safety profile. VT estimation via 2-TCM appears suitable for quantitative assessment of HDAC6 in the brain.Trial registrationJapan Registry of Clinical Trials (jRCT), jRCTs031240173. Registered 19 June 2024, https://jrct.mhlw.go.jp/latest-detail/jRCTs031240173.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12967-026-08449-8. Background Histone deacetylase 6 (HDAC6), a class IIb HDAC isoform, functions as an enzyme that catalyzes the deacetylation of lysine residues in various substrate proteins [1] and, despite its name, predominantly localizes to the cytoplasm, where it regulates multiple non-histone targets such α-tubulin [2]. In addition, HDAC6 contains distinct domains that bind ubiquitin and the dynein motor protein, facilitating transport of ubiquitinated proteins along microtubules [3]. Multiple studies have demonstrated that HDAC6 plays crucial roles in the pathogenesis of neurodegenerative diseases including Alzheimer’s disease and Parkinson’s disease [3–5]. Based on this body of evidence, HDAC6 has emerged as a promising therapeutic target for these disorders. Positron emission tomography (PET) enables noninvasive, quantitative imaging of biological processes in vivo. Postmortem studies have demonstrated altered HDAC6 expression in neurodegenerative brains [6, 7]; however, its regional and temporal dynamics during disease onset remain poorly characterized. HDAC6 PET imaging may delineate these patterns, clarify its pathological roles, and support the development of HDAC6-targeted therapies. [18F]EKZ-001 (also known as [18F]Bavarostat) is currently the most advanced PET radioligand for brain HDAC6 imaging and remains the only HDAC6-selective radioligand with reported clinical studies [8, 9]. [18F]FSW-100 is a potential HDAC6 PET radioligand derived from the HDAC6 inhibitor SW-100 developed by Kozikowski et al. [10, 11]. It employs a different 18F‑labelling chemistry compared with [18F]EKZ‑001. [18F]EKZ‑001 requires an uncommon ruthenium‑mediated deoxyradiofluorination reaction [12], and the ruthenium complex used in this reaction is generally synthesized in-house [13], making the reaction relatively demanding. The radiosynthesis of [18F]FSW-100 can be achieved via copper-mediated radiofluorination of arylboronic acid pinacol ester using the readily accessible copper catalyst Cu(OTf)2(py)4 [11]. Given the variation in available synthesis methods across facilities, offering multiple radioligand options could advance HDAC6 imaging research. According to a previous nonclinical validation study [14], [18F]FSW-100 exhibited high HDAC6 selectivity, with a half-maximal inhibitory concentration of 3 nM, and demonstrated good brain penetration and specific uptake in monkeys, as shown by blocking studies. Furthermore, toxicity, mutagenicity, and dosimetry data indicated an acceptable clinical safety profile [14]. In the present study, we performed the first-in-human PET study of [18F]FSW-100. The aims of this study were to evaluate the safety, biodistribution, and dosimetry of [18F]FSW-100 PET, and to investigate the optimal method for quantitative analysis in a brain PET study conducted with arterial blood sampling. Methods The study protocol was essentially the same as that reported previously [15, 16]. Study design This study was conducted in two cohorts. In cohort A, we estimated the biodistribution and whole-body dosimetry of [18F]FSW-100. In cohort B, we measured [18F]FSW-100 uptake in the brain and modelled radiotracer kinetics using an arterial input function corrected for radiometabolites. Safety monitoring was conducted in both cohorts. The study protocol (Trial ID: jRCTs031240173) was approved by the Ministry of Health, Labour and Welfare-certified Clinical Research Review Board of the Tokyo Metropolitan Institute for Geriatrics and Gerontology (CRB3180026), and was performed in accordance with the World Medical Association Declaration of Helsinki. Written informed consent was obtained from all volunteer subjects prior to the study. Study subjects and eligibility Subjects were recruited using the healthy volunteer database of 3 H Medi Solution (Tokyo, Japan). Cohorts A and B included three (36.0 ± 11.8 years old) and five (45.6 ± 6.5 years old) healthy men, respectively. For screening and anatomical coregistration, magnetic resonance imaging (MRI) was performed in cohort B using a Philips Ingenia Elition 3.0T scanner (Philips, Amsterdam, the Netherlands). In addition to the routine set of T2-weighted image, FLAIR image, Diffusion-weighted image, and Susceptibility-weighted image, a three-dimensional (3D) T1-weighted turbo field echo (repetition time, 6.7 ms; echo time, 3.9 ms; T1 delay, 700 ms; flip angle, 10°; matrix, 256 × 256 × 150 voxels) whole-brain image was acquired for each subject. The main exclusion criteria in cohorts A and B included any abnormal findings from physical or neurological examinations, or from clinical laboratory and physiological tests (including standard tests in biochemistry, hematology, and cardiovascular assessments such as electrocardiograms); current internal medical conditions (e.g., chronic anemia or chronic renal failure); a history of neurological or neuropsychiatric diseases; a history of severe drug hypersensitivity; current smoking; notable abnormalities on anatomical MRI; and any other condition deemed inappropriate for participation by the principal investigator. No specific pre-scan preparation, such as fasting, was requested prior to scanning. [18F]FSW-100 synthesis Radiosynthesis and quality control of [18F]FSW-100 were performed as described previously [14]. [18F]FSW-100 (n = 8) was obtained with a radiochemical purity of 97.2 ± 0.7% (range, 96.0–98.0%) and a molar activity of 1804 ± 618 GBq/µmol (range, 1033–2843 GBq/µmol) at the end of synthesis. The molar activity at the time of injection was 1099 ± 321 GBq/µmol (range, 634–1529 GBq/µmol). Subject dosing Subjects in cohort A received an average radioactivity dose of 185.2 ± 10.1 MBq (range, 176.1–196.1 MBq), and those in cohort B received 90.3 ± 8.9 MBq (range, 83.7–105.5 MBq). The average injected mass of FSW-100 was 0.06 ± 0.02 µg (range, 0.04–0.08 µg) for cohort A and 0.03 ± 0.01 µg (range, 0.02–0.04 µg) for cohort B. Table S1 (Additional file 1) lists the demographic data and dose information for each study cohort. Safety monitoring Vital signs, including heart rate, respiratory rate, pulse oximetry, blood pressure, and body temperature were monitored one week before and immediately before and after the PET scan. During the scan, respiratory rate, blood pressure, heart rate, and electrocardiogram were continuously monitored. Blood specimens were obtained one week before and one week after the injection of [18F]FSW-100 and subjected to blood biochemistry and hematology analyses. While the details of blood biochemistry and hematological test items have been reported previously [16], γ-glutamyl transpeptidase was newly included. Medical interviews were conducted by physicians one week before, on the day of, and one week after the scan. Subjects were contacted by phone or email the day after the scan to confirm the presence or absence of any subjective symptoms and objective abnormalities. Adverse events were recorded on the day of [18F]FSW-100 injection and throughout the one-week follow-up period. Whole-body imaging The study protocol for whole-body imaging and radiation dosimetry evaluation in human subjects was essentially the same as that reported previously [15, 16]. Whole-body PET/computed tomography (CT) was obtained using a Discovery MI PET/CT scanner (GE Healthcare, Milwaukee, WI) in 3D mode. The scanner provides a 20-cm axial field of view, spatial resolution of 3.9 mm full width at half maximum (FWHM), and z axis resolution of 4.50 mm FWHM [17]. [18F]FSW-100 was administered as bolus via a catheterized antecubital vein. Low-dose CT was used for attenuation and scatter correction of the PET emission scan. The first PET acquisition started 1 min after the start of intravenous injection of [18F]FSW-100. Whole-body imaging, covering the region from the top of the head to the midthigh, consisted of scans at 11 bed positions, with an overlap of 35 out of 71 slices per bed position. Frame settings for dynamic scans were 15 s/bed position × 4 frames, 30 s/bed position × 13 frames, and 60 s/bed position × 3 frames. Additional static images (120 s/bed position × 1 frame) were acquired at 4 and 6 h post-injection. After each scan, the subjects were allowed to leave the scanner bed, and urine was collected to measure both volume and radioactivity. Whole-body PET data were reconstructed under the following conditions: 3D ordered-subset expectation maximization (3D-OSEM, 34 subsets, 2 iterations); Gaussian post-filter, 5.0 mm (FWHM); 128 × 128 matrix, 431 slices; field of view, 700 mm; 5.47 × 5.47 × 2.79 mm/voxel. Volumes of interest (VOIs) were manually placed over six organs with high radioactivity uptake that could be identified from PET or the low-dose CT images: brain, gallbladder, small intestine, kidneys, liver, and urinary bladder. The decay-uncorrected and decay-corrected time–activity curves (TACs) of organs were calculated as the percentage injected dose (%ID) per organ. The normalized number of disintegrations (MBq-h/MBq administered) was then calculated for each source organ as the area under the uncorrected TAC multiplied by the volume of the organ VOI. The area under the TAC was obtained by combining the area from time zero to the endpoint (by trapezoidal integration) and the area from the endpoint to infinity (by integrating radioactive decay beyond the endpoint). The absorbed dose in 25 target organs of an adult male phantom was estimated from the normalized number of disintegrations of source organs by implementing the MIRD method using OLINDA/EXM (Vanderbilt University, Nashville, TN) [18]. The effective dose was calculated using the methodology described in the Recommendations of the International Commission on Radiological Protection (Publication 103) [19]. To consider the influence of urination, we also calculated doses under the condition in which the contribution of excreted urine was excluded from the normalized number of disintegrations in the “remainder body”, termed the “with urination” condition. Brain imaging The study protocols for brain imaging and kinetic analysis in human subjects were essentially the same as those reported previously [15, 16]. Brain PET/CT was obtained using a Discovery MI PET/CT scanner. Following a low-dose CT scan for attenuation and scatter correction, [18F]FSW-100 was administered slowly via a catheterized antecubital vein over exactly 1 min using an automatic pump (Auto enhance A-25, Nemoto Kyorindo, Tokyo, Japan), and a 120-min dynamic scan was subsequently performed in 3D mode (20 s × 3 frames, 30 s × 3 frames, 60 s × 5 frames, 150 s × 5 frames, and 300 s × 20 frames). Arterial blood samples (0.5 mL each) were collected via a catheter placed in the radial artery at 10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 110, 120, 135, 150, and 180 s and 5, 7, 10,15, 20, 30, 40, 50, 60, 75, 90, 105, and 120 min. The whole blood and separated plasma were weighed, and radioactivity was measured with a NaI (Tl) well scintillation counter (BeWell Model-QS03 F/B; Alpha Systems, Wakayama, Japan). To analyze radiometabolites, an additional 1.5 mL of blood was collected at 3, 10, 20, 30, 40, 60 and 90 min after dosing. Prior to the PET scan, approximately 4 mL of arterial blood was collected to determine the plasma free fraction (fP) using the previously described ultrafiltration method [20]. After the PET scan, urine was collected to determine both volume and radioactivity, and was subjected to metabolite analysis. Metabolite analysis Radiometabolite analyses of plasma and urine samples were performed using high-performance liquid chromatography (HPLC), largely based on previously published methods [15, 16]. Plasma fractions were separated by centrifugation at 8700 × g for 40 s at 2 °C from the arterial blood samples collected during brain PET imaging. The resulting plasma was kept chilled on crushed ice and mixed with an equal volume of acetonitrile, followed by centrifugation at 8700 × g for 30 s at 2 °C. The radioactivity of both the supernatant and the precipitate was measured using a Hidex Automatic Gamma Counter (Hidex, Turku, Finland). The extraction efficiency of radioactivity from the samples at each time point was 84%–89%. The supernatant was then centrifuged again under the same conditions for 5 min to remove any residual insoluble material and to prevent clogging of the HPLC system, and the final supernatant was analyzed by HPLC (column: Sunniest C18, 5 μm, 10 × 250 mm; eluent: acetonitrile/water/formic acid = 60/40/0.1; flow rate: 4.5 mL/min). The retention time of FSW-100 was 5.6 min. Fractions were collected at 0.3-min intervals, and their radioactivity was measured using the gamma counter. The mean radioactivity recovery yield from HPLC at each time point ranged from 84% to 111%. A 1-mL aliquot of urine from each subject in the brain PET imaging study was centrifuged at 8700 × g for 30 s at 2 °C. The resulting supernatant was centrifuged again under the same conditions for 5 min. These centrifugations were performed to remove residual insoluble materials and to prevent clogging of the HPLC system, although no visible precipitate was observed. Radiometabolites present in the final supernatant were analyzed by HPLC, following the method described above. Brain image and kinetic analysis Brain PET data were reconstructed under the following conditions: 3D-OSEM (16 subsets, 4 iterations) with incorporation of time-of-flight information; Gaussian post-filter, 4.0 mm (FWHM); 128 × 128 matrix, 71 slices; field of view, 256 mm; 2.00 × 2.00 × 2.79 mm/voxel. Using the MRI obtained for coregistration as a reference, VOIs defined as a set of partially overlapped circles (10 mm in diameter) were placed in 10 brain regions: the frontal, temporal and cerebellar cortices, thalamus, putamen, caudate nucleus, anterior and posterior cingulate gyri, hippocampus, and white matter (centrum semiovale). TACs for these VOIs were calculated in units of Bq/mL or expressed as standardized uptake value (SUV), defined as (activity/mL tissue)/(injected activity/body weight). The time course of the unchanged fraction in plasma was modeled using a tri-exponential function, and the metabolite-corrected arterial plasma TAC was calculated as the product of the plasma activities and the interpolated or extrapolated unchanged fraction. Prior to kinetic modeling, the relative transport delay between the averaged TAC of all VOIs and the arterial TAC function was estimated from early-phase TACs (0–20 min) and corrected in subsequent analyses. The distribution volume (VT) of [18F]FSW-100 was estimated by applying both a one-tissue-compartment model (1-TCM; K1/k2) and a two-tissue-compartment model (2-TCM; K1/k2 × (1 + k3/k4)). The goodness of fit between the two model analyses was evaluated using the Akaike information criterion (AIC). VT was also estimated by the Logan graphical analysis (LGA; Logan plot) (t* = 20 min) [21, 22], and parametric VT images were estimated by voxel-wise LGA. To examine the impact of inter-individual variability on the unchanged fraction, we investigated 2-TCM using the averaged (n = 5) unchanged fraction for the calculation of plasma TAC. For further evaluation of quantification, the distribution volume ratio (DVR) was calculated using white matter, characterized by markedly lower radioligand accumulation compared to other regions, as the candidate reference region. The DVRs were calculated using the VT values estimated by the 2-TCM, in which the individual unchanged fractions were incorporated. The impact of scan duration on regional VT values was evaluated by calculating bias, expressed as the percentage difference between shortened scan durations and the full 120-min acquisition, using VT estimates derived from five subjects, each with ten VOIs. To facilitate comparison with previously published VT data for [18F]EKZ-001 [8], regional VTs of [18F]FSW-100 were calculated using TACs extracted from VOIs defined by the Hammers atlas [23], thereby aligning the quantification methodology as closely as possible. These evaluations were performed using PMOD software (Ver. 4.3, Bruker, Billerica, MA) and in-house MATLAB-based (MathWorks, Natick, MA) programs. Results Safety monitoring The scan protocol was well tolerated by all subjects in cohorts A and B. No adverse events or clinically detectable pharmacologic effects were observed in any subject. No clinically meaningful trends suggestive of a safety signal were observed in any laboratory parameter, vital sign, or electrocardiogram. Whole-body biodistribution and dosimetry After the initial 2-h whole-body scan, the mean decay-corrected radioactivity recovered in urine was 12%ID (Additional file 1: Table S2, n = 8). The cumulative recovery of radioactivity in urine at 6 h post-injection was 23%ID (n = 3). Figure 1 presents whole-body [18F]FSW-100 PET images of one subject, and Fig. 2 presents mean regional TACs. Immediately after [18F]FSW-100 administration, relatively high accumulation of radioactivity was observed in the brain, liver, kidneys, and small intestine. Radioactivity in these organs decreased over time; however, in the small intestine, it began increasing again approximately 90 min after administration. Radioactivity in the gallbladder and urinary bladder increased over time, reaching peak levels at approximately 120 min before subsequently decreasing. The mean normalized number of disintegrations in the source organs are summarized in Table S3 (Additional file 1). Fig. 1Whole-body maximum-intensity-projection [18F]FSW-100 PET images for a subject in cohort A. PET image intensities are decay-corrected and expressed as SUV. The numbers below the images are time elapsed since administration of [18F]FSW-100 Whole-body maximum-intensity-projection [18F]FSW-100 PET images for a subject in cohort A. PET image intensities are decay-corrected and expressed as SUV. The numbers below the images are time elapsed since administration of [18F]FSW-100 Fig. 2Regional decay-corrected TACs of six source organs after administration of [18F]FSW-100. Data are expressed as mean (n = 3) Regional decay-corrected TACs of six source organs after administration of [18F]FSW-100. Data are expressed as mean (n = 3) Table 1 summarizes organ absorbed dose and effective dose, estimated under two conditions; with and without urination. The gallbladder wall exhibited the highest absorbed dose of 379 ± 136 (range, 273–533) µGy/MBq, irrespective of urination status. The whole-body effective dose was estimated at 18.2 ± 1.6 (range, 16.5–19.7) and 18.9 ± 1.7 (range, 17.0–20.4) µSv/MBq with and without urination, respectively. Table 1Organ absorbed doses and effective dose of [18F]FSW-100OrganAbsorbed dose (µGy/MBq)Without urinationWith urinationAdrenals14.03 ± 0.6113.20 ± 0.75Brain31.30 ± 8.5531.17 ± 8.50Breasts7.11 ± 0.626.46 ± 0.78Gallbladder wall379.33 ± 136.31379.00 ± 136.48Lower large intestine wall14.50 ± 0.7913.57 ± 0.85Small intestine57.63 ± 1.5956.73 ± 1.65Stomach wall12.50 ± 0.5611.60 ± 0.75Upper large intestine wall22.57 ± 0.8121.73 ± 0.91Heart wall10.63 ± 0.599.76 ± 0.85Kidneys35.43 ± 7.9235.03 ± 7.90Liver43.80 ± 7.8943.47 ± 7.83Lungs9.26 ± 0.568.50 ± 0.76Muscle9.84 ± 0.499.07 ± 0.67Ovaries16.70 ± 0.7215.77 ± 0.80Pancreas16.50 ± 1.2315.57 ± 1.32Red marrow10.18 ± 0.369.48 ± 0.55Osteogenic cells13.60 ± 1.0612.40 ± 1.30Skin6.98 ± 0.536.37 ± 0.66Spleen10.60 ± 0.629.75 ± 0.88Testes8.95 ± 0.708.17 ± 0.80Thymus8.37 ± 0.877.55 ± 1.05Thyroid8.41 ± 0.817.61 ± 0.96Urinary bladder wall79.53 ± 22.0378.70 ± 21.88Uterus18.60 ± 1.0117.73 ± 0.80Total body11.87 ± 0.1211.10 ± 0.30Effective dose (µSv/MBq)18.88 ± 1.7018.18 ± 1.60Data are expressed as mean ± standard deviation (n = 3) Table 1Organ absorbed doses and effective dose of [18F]FSW-100OrganAbsorbed dose (µGy/MBq)Without urinationWith urinationAdrenals14.03 ± 0.6113.20 ± 0.75Brain31.30 ± 8.5531.17 ± 8.50Breasts7.11 ± 0.626.46 ± 0.78Gallbladder wall379.33 ± 136.31379.00 ± 136.48Lower large intestine wall14.50 ± 0.7913.57 ± 0.85Small intestine57.63 ± 1.5956.73 ± 1.65Stomach wall12.50 ± 0.5611.60 ± 0.75Upper large intestine wall22.57 ± 0.8121.73 ± 0.91Heart wall10.63 ± 0.599.76 ± 0.85Kidneys35.43 ± 7.9235.03 ± 7.90Liver43.80 ± 7.8943.47 ± 7.83Lungs9.26 ± 0.568.50 ± 0.76Muscle9.84 ± 0.499.07 ± 0.67Ovaries16.70 ± 0.7215.77 ± 0.80Pancreas16.50 ± 1.2315.57 ± 1.32Red marrow10.18 ± 0.369.48 ± 0.55Osteogenic cells13.60 ± 1.0612.40 ± 1.30Skin6.98 ± 0.536.37 ± 0.66Spleen10.60 ± 0.629.75 ± 0.88Testes8.95 ± 0.708.17 ± 0.80Thymus8.37 ± 0.877.55 ± 1.05Thyroid8.41 ± 0.817.61 ± 0.96Urinary bladder wall79.53 ± 22.0378.70 ± 21.88Uterus18.60 ± 1.0117.73 ± 0.80Total body11.87 ± 0.1211.10 ± 0.30Effective dose (µSv/MBq)18.88 ± 1.7018.18 ± 1.60Data are expressed as mean ± standard deviation (n = 3) Organ absorbed doses and effective dose of [18F]FSW-100 Data are expressed as mean ± standard deviation (n = 3) Plasma free fraction and metabolite analysis The mean fP was 2.7 ± 0.3%. Figure 3 illustrates TACs for blood total radioactivity, plasma total radioactivity, and plasma radioactivity corrected for the individual unchanged fraction, and the time course of the unchanged fraction. The plasma radioactivity decreased immediately following [18F]FSW-100 administration and remained relatively stable from 10 min post-administration onward. The unchanged fraction also decreased rapidly following [18F]FSW-100 administration, reaching 23 ± 8% at 10 min post-administration before gradually decreasing over time; however, after 60 min post-administration, an increase in the peak near the retention time of the unchanged form was observed in HPLC analysis, raising suspicion of interference from radiometabolites (Additional file 1: Figs. S1 and S2). Therefore, metabolite correction of the arterial input function was performed using the unchanged fraction data obtained within 40 min post-administration. HPLC analysis of plasma revealed polar radiometabolites and a carboxylic acid form of [18F]FSW-100 (Additional file 1: Fig. S1 and Table S4), consistent with the metabolic patterns observed in mice and monkeys [11, 14]. Radioactivity chromatograms of urine samples indicated the absence of both the unchanged compound and the carboxylic acid form, with only polar radiometabolites detected (Additional file 1: Fig. S1(i)). Fig. 3Blood and plasma radioactivity TACs and unchanged fraction. TACs for blood total radioactivity, plasma total radioactivity, and plasma radioactivity corrected for the unchanged fraction (a); and time course of the unchanged fraction in plasma (b). An enlarged view of the 0–10 min post-injection period is inserted in panel a Blood and plasma radioactivity TACs and unchanged fraction. TACs for blood total radioactivity, plasma total radioactivity, and plasma radioactivity corrected for the unchanged fraction (a); and time course of the unchanged fraction in plasma (b). An enlarged view of the 0–10 min post-injection period is inserted in panel a Brain PET imaging and kinetic modeling Representative brain MRI and PET images (static SUV and LGA VT images) are shown in Fig. 4, and the mean TACs are presented in Fig. 5. In the gray matter regions, regional differences in the TACs of [18F]FSW-100 were minimal, and following administration of [18F]FSW-100, radioactivity rapidly increased and then gradually washed out. In the hippocampus, moderate early-phase uptake was observed followed by a relatively stable level thereafter. In the white matter, initial uptake was low, and little change was observed over time. Regional VT values estimated by 1-TCM, 2-TCM, and LGA are presented in Table 2. The highest VT values were observed in the cerebellar gray matter and putamen, with the lowest value found in the white matter. Comparison of model-fitting using the AIC across all investigated regions showed that 2-TCM yielded lower AIC scores than 1-TCM (Additional file 1: Fig. S3), suggesting that the former is a more suitable model for quantitative analysis of [18F]FSW-100. VT values derived from 2-TCM showed a strong correlation with those estimated by LGA. However, the LGA slightly underestimated the values by approximately 5%, and a proportional bias was observed (Additional file 1: Fig. S4). Table 2 also presents the regional VT values estimated by 2-TCM using the averaged unchanged fraction, along with the DVRs calculated using white matter as the reference region. The coefficients of variation (%CVs) for both the VT values and DVRs obtained using these methods were lower than those of the original VT values. The impact of scan duration on regional VT values is shown in Fig. S5 (Additional file 1). Compared to the 120-min scans, VT estimates from 90-min scans were acceptable, showing a mean bias of less than 5% for both 2-TCM and LGA. Table 2Estimated regional VT values (mL/cm3) and DVR of [18F]FSW-100 in the brainBrain regionVT (1-TCM)VT (2-TCM)VT (LGA)VT (2-TCM)DVR (white matter as reference region)(unchanged fraction)individualindividualindividualaveragedindividualCerebellar gray matter41.9 ± 9.7 (23.1)42.6 ± 9.7 (22.8)41.5 ± 9.5 (22.8)45.3 ± 7.9 (17.4)3.6 ± 0.4 (11.3)Frontal cortex32.9 ± 9.3 (28.3)33.2 ± 9.2 (27.6)32.2 ± 8.9 (27.5)34.7 ± 6.2 (18.0)2.7 ± 0.3 (9.4)Temporal cortex37.1 ± 9.9 (26.7)37.8 ± 9.7 (25.8)36.4 ± 9.3 (25.6)39.9 ± 7.4 (18.4)3.1 ± 0.2 (6.8)Thalamus30.9 ± 7.7 (25.0)30.8 ± 7.5 (24.4)30.0 ± 7.3 (24.3)32.3 ± 6.7 (20.8)2.6 ± 0.3 (9.9)Putamen40.0 ± 10.5 (26.3)40.1 ± 9.9 (24.6)38.7 ± 9.5 (24.5)42.4 ± 8.2 (19.4)3.3 ± 0.2 (7.1)Caudate nucleus33.0 ± 8.9 (27.1)33.6 ± 8.7 (26.0)32.5 ± 8.2 (25.3)35.3 ± 6.0 (16.9)2.8 ± 0.2 (8.7)Hippocampus31.9 ± 10.0 (31.3)35.0 ± 10.7 (30.5)33.0 ± 9.7 (29.4)37.5 ± 9.1 (24.3)2.9 ± 0.1 (5.0)Anterior cingulate gyrus34.5 ± 9.6 (27.9)35.4 ± 9.6 (27.0)33.8 ± 9.4 (27.8)37.1 ± 6.2 (16.7)2.9 ± 0.3 (10.7)Posterior cingulate gyrus34.6 ± 8.1 (23.5)34.6 ± 8.0 (23.2)33.5 ± 7.6 (22.6)36.2 ± 5.5 (15.1)2.9 ± 0.3 (10.4)White matter11.0 ± 3.1 (27.9)12.2 ± 3.6 (29.2)12.1 ± 3.6 (29.4)13.0 ± 3.3 (25.6)–Data are expressed as mean ± standard deviation (%CV) (n = 5) Table 2Estimated regional VT values (mL/cm3) and DVR of [18F]FSW-100 in the brainBrain regionVT (1-TCM)VT (2-TCM)VT (LGA)VT (2-TCM)DVR (white matter as reference region)(unchanged fraction)individualindividualindividualaveragedindividualCerebellar gray matter41.9 ± 9.7 (23.1)42.6 ± 9.7 (22.8)41.5 ± 9.5 (22.8)45.3 ± 7.9 (17.4)3.6 ± 0.4 (11.3)Frontal cortex32.9 ± 9.3 (28.3)33.2 ± 9.2 (27.6)32.2 ± 8.9 (27.5)34.7 ± 6.2 (18.0)2.7 ± 0.3 (9.4)Temporal cortex37.1 ± 9.9 (26.7)37.8 ± 9.7 (25.8)36.4 ± 9.3 (25.6)39.9 ± 7.4 (18.4)3.1 ± 0.2 (6.8)Thalamus30.9 ± 7.7 (25.0)30.8 ± 7.5 (24.4)30.0 ± 7.3 (24.3)32.3 ± 6.7 (20.8)2.6 ± 0.3 (9.9)Putamen40.0 ± 10.5 (26.3)40.1 ± 9.9 (24.6)38.7 ± 9.5 (24.5)42.4 ± 8.2 (19.4)3.3 ± 0.2 (7.1)Caudate nucleus33.0 ± 8.9 (27.1)33.6 ± 8.7 (26.0)32.5 ± 8.2 (25.3)35.3 ± 6.0 (16.9)2.8 ± 0.2 (8.7)Hippocampus31.9 ± 10.0 (31.3)35.0 ± 10.7 (30.5)33.0 ± 9.7 (29.4)37.5 ± 9.1 (24.3)2.9 ± 0.1 (5.0)Anterior cingulate gyrus34.5 ± 9.6 (27.9)35.4 ± 9.6 (27.0)33.8 ± 9.4 (27.8)37.1 ± 6.2 (16.7)2.9 ± 0.3 (10.7)Posterior cingulate gyrus34.6 ± 8.1 (23.5)34.6 ± 8.0 (23.2)33.5 ± 7.6 (22.6)36.2 ± 5.5 (15.1)2.9 ± 0.3 (10.4)White matter11.0 ± 3.1 (27.9)12.2 ± 3.6 (29.2)12.1 ± 3.6 (29.4)13.0 ± 3.3 (25.6)–Data are expressed as mean ± standard deviation (%CV) (n = 5) Estimated regional VT values (mL/cm3) and DVR of [18F]FSW-100 in the brain Data are expressed as mean ± standard deviation (%CV) (n = 5) Fig. 4Brain MRI, SUV images, and VT maps. Representative brain MRI, [18F]FSW-100 SUV images (summed from 60 to 90 min after administration), and LGA VT parametric maps Brain MRI, SUV images, and VT maps. Representative brain MRI, [18F]FSW-100 SUV images (summed from 60 to 90 min after administration), and LGA VT parametric maps Fig. 5Time–activity curves in seven brain regions after [18F]FSW-100 administration. Data are expressed as the mean (n = 5) Time–activity curves in seven brain regions after [18F]FSW-100 administration. Data are expressed as the mean (n = 5) Discussion This paper reports the first-in-human PET study of [18F]FSW-100 for imaging HDAC6 in the brain. As an initial phase of clinical research, this study aimed to evaluate the safety, whole-body biodistribution, and radiation dosimetry of [18F]FSW-100, and also to explore optimal kinetic modeling using dynamic brain PET. From a safety standpoint, the [18F]FSW-100 PET scan was well tolerated by healthy participants. In a previous nonclinical acute toxicity study, no clinical signs of toxicity were observed in rats that received intraperitoneal administration of FSW-100 at a dose of 2.5 mg/kg [14]. In the present study, the amount of FSW-100 administered to subjects in cohort A was 0.06 ± 0.02 µg. This corresponds to 1.0 ± 0.4 ng/kg, which is 1/2,500,000 (1/405,000 when allometric scaling is applied [24]) of the dose administered in the rat toxicity study, indicating a sufficient safety margin. Whole-body [18F]FSW-100 PET imaging was performed to assess human radiation dosimetry. After [18F]FSW-100 administration, radioactivity accumulated in the gallbladder and urinary bladder, suggesting dual elimination via renal and hepatobiliary routes. Within six hours, 23%ID was excreted in human urine, whereas mice showed 75%ID, indicating that mice have a faster urinary excretion rate than humans [14]. In mice, 70% of plasma radioactivity originated from the carboxylic acid metabolite within 15 min, with only 6% remaining as unchanged [18F]FSW-100 [11]. In contrast, [18F]FSW-100 metabolism in humans was relatively slower, with 14% unchanged and 22% carboxylic acid metabolite at 20 min (Additional file 1: Table S4). These species-specific metabolic differences likely affect their respective excretion pathways. The effective dose estimated from human biodistribution was 18.9 ± 1.7 µSv/MBq, representing 111% of the mouse-derived value [14] and falling within the range reported for existing 18F-labelled PET radioligands [25]. When 185 MBq of [18F]FSW-100 was administered, the effective dose was estimated at 3.49 mSv, which is below the limit of 10 mSv recommended for clinical research by the ICRP, and practiced in Europe [26, 27]. In contrast, in the gallbladder wall, a 185 MBq dose would result in an absorbed dose of 99 mGy, based on the highest value in cohort A. This exceeds the single-study limits for individual organs (30 mGy for sensitive organs, 50 mGy for others) recommended by the US Radioactive Drug Research Committee (RDRC) [28]. For clarification, this threshold serves as one of the criteria for determining whether a clinical study falls within the scope of review by the RDRC [29]. While medical exposures are not subject to dose limits, radiation doses should still be minimized to the extent that diagnostic or research objectives are not compromised [30]. In line with this principle, and in the absence of established standards or guidelines for radiation protection in research volunteers in Japan, the present study implemented precautionary measures by administering to cohort B half the dose given to cohort A, based on the estimated absorbed dose to the gallbladder wall. Regarding the accumulation of radioactivity in the gallbladder, it remains unclear whether the observed activity represents the unchanged form or radiometabolites. In general, brain imaging radioligands are highly lipophilic, and such compounds tend to undergo hepatobiliary excretion, which often results in the gallbladder becoming the critical organ [25]. In the present study, the normalized number of disintegrations in the gallbladder showed substantial inter-subject variability (Table S3). Therefore, from the perspective of radiation protection of research subjects, it is important to consider inter-individual variability in gallbladder radiation dose and to control factors that influence gallbladder physiology. Figure S6 (Additional file 1) shows the individual TACs for the gallbladder. Notably, subject No. 1 showed a markedly different activity profile: gallbladder activity remained elevated after the peak, whereas the other subjects showed a gradual decline over time. This difference was likely attributable to the absence of specific instructions regarding food intake before and after imaging. A previous study using a vesicular acetylcholine transporter radioligand demonstrated that caloric intake, either before or after imaging, reduced the radiation dose to the gallbladder compared to fasting conditions [31]. Therefore, future studies should consider implementing a protocol that includes food or drink intake to promote gallbladder contraction before or after [18F]FSW-100 PET imaging [32]. Dynamic brain imaging data were used to identify the optimal model for quantitative analysis of [18F]FSW-100. Relatively high VT values were observed in the cerebellum and putamen; however, regional differences in the VT values were generally small, except in the white matter (Table 2). While current understanding of HDAC6 expression and distribution within the brain remains limited, its mRNA expression exhibits low regional specificity across brain areas [33]. Although manual VOI placement was typically employed in first-in-human studies conducted at our institution, VT values for [18F]FSW-100 were also derived using Hammers-atlas-based VOIs [23] to enable standardized regional comparisons. These values were slightly lower than those obtained using manually defined VOIs, particularly in the cerebral and cerebellar cortex. This discrepancy is likely due to the manual VOIs being selected from regions exhibiting high uptake during the early phase, aiming to extract representative signals and minimize heterogeneity (Additional file 1: Fig. S7). When compared to the representative HDAC6 radioligand [18F]EKZ-001, [18F]FSW-100 showed slightly lower VT values overall, with a moderate reduction observed in the hippocampus and brainstem (Additional file 1: Fig. S8) [8]. It may be too early to discuss inter-individual variability given the small sample size; nevertheless, as a challenge in the quantitative analysis of [18F]FSW-100, considerable variability in the estimated VT values was observed (2-TCM; %CV > 20%). In contrast, the variability in fP values of [18F]FSW-100 was low. The mean %CVs of VT and VT/fP across individual brain regions were 26 ± 3% and 27 ± 4%, respectively. Therefore, correction of VT using fP did not significantly affect the level of variability. While variability in [18F]FSW-100 VT may partly reflect individual differences in HDAC6 expression, instability in the input function, particularly in the unchanged fraction, is likely a major contributing factor. As shown in Fig. 3b, [18F]FSW-100 exhibited poor plasma stability. Furthermore, in the later scan phase, accurate quantification of the unchanged fraction was compromised by co-elution of radiolabeled metabolites with the parent compound in HPLC analysis and by reduced radioactivity due to decay. Notably, when the averaged unchanged fraction was used, the %CV of regional VT values was 19 ± 3%, representing a 26 ± 9% reduction compared to the original estimates (Table 2), suggesting that instability in the unchanged fraction is a key driver of VT variability. As an additional evaluation of [18F]FSW-100 quantification methods, DVR values calculated using white matter as the candidate reference region showed relatively low inter-individual variability (9 ± 2%), suggesting that DVR may be effective in reducing variability. However, it remains necessary to confirm whether white matter is a suitable reference region. If the use of a reference region is validated, approaches such as reference LGA [34] could eliminate the need for arterial blood sampling. One potential confounding factor in the quantification analysis is possible brain entry of radiometabolites; however, the major metabolite of [18F]FSW-100, the carboxylic acid form, shows limited brain penetration in mice [11]. While dual-input models may serve as an alternative analytical strategy [35], the available information on the low brain penetration of this metabolite led us not to adopt this approach in the present study. Overall, 2-TCM and LGA appear preferable for estimating regional VT values of [18F]FSW-100, but further clinical studies are needed to establish a robust quantification approach. Given that future studies may involve elderly subjects or patients, a shorter scan duration should be considered; based on current comparisons (Additional file 1: Fig. S5), a 90-min protocol may be sufficient. In addition, improving metabolic analysis methods may be warranted. Although challenging, enhancing radioligand stability through chemical modification could also be considered. This study has several limitations, such as a small sample size and the absence of a test–retest reproducibility assessment. Among these, a major limitation is that the study was conducted only in men. According to the Standards for PET Drugs Used in Molecular Imaging Clinical Research established by the Japanese Society of Nuclear Medicine [36], exploratory clinical studies designed to evaluate safety and radiation dosimetry should, in principle, be conducted in adult men, because additional ethical and safety considerations are required for women of childbearing potential. In line with this guidance, the inclusion of women was considered appropriate only after confirming the safety and pharmacokinetic profile of [18F]FSW-100. However, a sex difference in HDAC6 brain levels, with men showing higher expression than women, has been reported using [18F]EKZ-001 [8]. Given the importance of HDAC6 in conditions known to exhibit pronounced sex differences, including neurodegenerative and autoimmune diseases [37–39], it will be essential to include women in future studies. Importantly, we are currently planning a test–retest clinical study of [18F]FSW-100, in which we will recruit an equal number of male and female participants to appropriately address potential sex differences. Another limitation of this study is that metallo-beta-lactamase domain-containing protein 2 (MBLAC2) has been identified as an off-target of various HDAC6 inhibitors [40]. Accurate interpretation of [18F]FSW-100 PET data therefore requires evaluation of MBLAC2 expression in the brain and determining its contribution to radioligand accumulation via off-target binding. Conclusions In this first-in-human PET study, [18F]FSW-100 PET scans were conducted safely. The estimated effective dose was within acceptable limits; however, a high absorbed dose in the gallbladder wall necessitated caution regarding the administered dosage. Quantitative analysis of brain uptake was feasible using 2-TCM or LGA, although considerable inter-individual variability was observed, likely due to low metabolic stability of the radioligand. Further validation of [18F]FSW-100 is needed through studies with larger cohorts, test–retest evaluations, and assessment of sex-related differences. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1:This file contains Supplementary Figures (Figure S1–S8) and Supplementary Tables (Table S1–S4). Supplementary Material 1:This file contains Supplementary Figures (Figure S1–S8) and Supplementary Tables (Table S1–S4).