No-carrier-added aqueous [18F]fluoride was produced according to previously described procedures [10]. Briefly, an ion-exchange resin (Sep-Pak Accell Light QMA cartridge, Waters Corporation, Milford, MA, USA) was prepared by washing with NaCl (0.7 mL, 0.9 mg/mL) in water (9.3 mL). Next, aqueous [18F]fluoride was drawn into a collection syringe and passed through the anion-exchange resin in the carbonate form. The trapped [18F]fluoride was rinsed with sterile water (10 mL) to remove contaminants and traces of irradiated water. Finally, [18F]NaF was eluted from the ion-exchange cartridge with a NaCl-solution (10 mL, 9 mg/mL). The [18F]NaF was then formulated for injection using a sterile filtration unit combined with the synthesis device. The end product was filtered through a sterile filter (Millex GP 0.22 µm, EDM Millipore Billerica, MA, USA), and placed into a sterile, pyrogen-free, vial.

We used Harlan Sprague–Dawley rats (n = 39, 21 males and 18 females), bred and housed at the Central Animal Laboratory, University of Turku, Turku, Finland. The males were younger (55 ± 16 d) than the females (112 ± 6 d), but the body weights (male; 286 ± 8 g, and female; 260 ± 25 g) were not significantly different between the sexes. All rats were housed under standard conditions (temperature 21 °C; humidity 55 ± 5%) with lights on from 6:00 a.m. to 6 p.m. The rats had free access to standard laboratory food and tap water. All animal experiments were approved by the Animal Experiment Board of the Province of Southern Finland (ESAVI/4660/04.10.07/2016), and carried out according to ARRIVE guidelines; the United Kingdom Animals (Scientific Procedures) Act, 1986, and EU Directive 2010/63/EU for animal experiments.

[18F]NaF was injected intravenously via a tail vein into male (n = 15, 32.3 ± 6.4 MBq) and female rats (n = 18, 32.2 ± 4.5 MBq) under brief isoflurane/oxygen anaesthesia. The rats (n = 6 for each time point) were sacrificed via cardiac puncture under increased anaesthesia at 15, 30, 60, 120, 240, or 360 min after injecting the tracer. Blood, urine, and organs of interest were removed, weighed, and measured for 18F-radioactivity in a NaI(Tl) well counter (3 × 3-inch, Bicron, Newbury, USA). The uptake of 18F-radioactivity was expressed as the percentage of the injected dose per gram of tissue (% inj.dose/g) or as the organ-to-blood ratio. The binding of [18F]fluoride to plasma proteins was determined by ultrafiltration, as previously described [11]. We used a haematocrit value of 41% [12] for calculating the portion of [18F]fluoride that bound to erythrocytes.

The total radioactivity in the urine was calculated by pooling the amount of radioactivity in the bladder with the amount excreted in the urine throughout the study.

Three male rats (two of them twice) (272 ± 26 g) were anesthetized with an isoflurane/oxygen gas mixture. [18F]NaF was intravenously injected via a tail vein (18.7 ± 2.4 MBq). Rats were scanned with an Inveon multimodality PET/CT scanner (Siemens Medical Solutions, Knoxville, TN). Transmission scans with the CT modality were first performed for anatomical reference and for correcting attenuations of PET data. Dynamic PET scans were acquired in list mode with a 350–650 keV energy window. Scans were initiated at the same time that radiotracer was injected and continued for 60 min. Sinograms were framed into 53 time frames (4 × 5 s, 28 × 10 s, 15 × 60 s, 4 × 300 s, and 2 × 600 s) and reconstructed with Fourier rebinning and a two-dimensional filtered backprojection reconstruction algorithm. Volumes of interest (VOIs) were drawn over compact (cortical) bones, cancellous (trabecular) bones, and flat bones. Compact bone samples were taken from the diaphysis of long bones (tibia and radius). Cancellous bone samples were taken from the epiphyseal regions of long bones (tibia head and radial head). In addition to bone samples, VOIs were drawn over the following soft tissues; whole brain, cardiac left ventricle, liver, kidneys, bone marrow, and bladder, with Inveon Research Workplace Image Analysis software (Siemens Medical Solutions). The image voxel size was 0.86 mm × 0.86 mm × 0.80 mm. From the VOIs, time–activity curves (TACs) were obtained. [18F]fluoride uptake was expressed as the percentage of the injected dose per millilitre of tissue (% inj.dose/mL).

For kinetic analyses, we used the whole blood TAC, measured from the left cardiac ventricle as the input function, in images where the heart was in the field of view (n = 3 rats). Fluoride is rapidly transported through the red blood cell membrane; thus, [18F]fluoride in whole blood is available for exchange in tissue capillaries [2, 13]. Blood TACs could be reliably determined from the image, because the ventricle was relatively large compared to the image resolution, and the activity ratio between myocardium and blood remained stable during the PET study (Additional file 1: Table S2). The net [18F]fluoride influx rate (Ki) in bone was assessed with a Patlak graphical analysis. The rate constants for [18F]fluoride transport from blood to the extravascular compartment (K1) and back to the blood (k2), the rate constants for binding to and detaching from hydroxyapatite at the bone surface (k3 and k4), and the vascular volume fraction (Vb) were parameters of the three-compartmental model [14,15,16] (Additional file 1: Fig. S1). The rate constant, K1, mainly represented [18F]fluoride perfusion. The ratio Ki/K1 (in terms of the compartmental model: k3/(k2 + k3), was the unidirectional extraction efficiency from blood to bone minerals [17]. Detachment of fluoride-18 from bone mineral was assumed negligible during the PET study (k4 = 0).

The statistical analyses were performed using the SAS software, version 9.4 for Windows (SAS Institute Inc., Cary, NC, USA). The K1, Ki, K1/Ki were compared for more bone in the same rat, therefore the analysis was done using repeated measures techniques, where bone is a repeated factor (hierarchical linear mixed model). While overall differences were detected between the bones, pairwise comparisons were made between the bones. Urinary excretion of radioactivity in females and males was analysed using two-way analysis of variance (ANOVA), using time and gender as explanatory variables (at each time point different rat was measured). Values are expressed as the means ± standard deviation (SD). P values < 0.05 (two-tailed) were considered significant.