The results of this study were relevant to both clinical and experimental settings. First, our findings on the pharmacokinetic properties of [18F]fluoride in different bones as a function of time were relevant for adapting tracer protocols to fit different clinical applications, and when assessing the bone uptake of 18F-labelled tracers that release [18F]fluoride. Second, our findings on the behaviour of free [18F]fluoride in soft tissues improved our understanding of when defluorination occurs in the metabolism of 18F-labelled tracers; this information might have an impact on future kinetic modeling approaches.
It is well known that [18F]fluoride is taken up by bones; for example, the rate of [18F]fluoride influx into bone mineral (Ki) was shown to be irreversible by using a Patlak graphical analysis [3]. We studied the uptake of [18F]fluoride into various rat bones with both in vivo PET imaging (for 60 min after tracer injection) and ex vivo gamma counting (for 360 min after tracer injection). We demonstrated a good correlation between these measurements at 60 min post injection. With both methodologies, the highest uptake was observed in the epiphyseal region of tibia. However, the uptake in the epiphyseal regions of the tibia appeared to be more related to bone perfusion (K1) than to bone uptake (k3). Ki accurately describes the net influx of fluoride in tissues, but cannot separate the effect of perfusion from fluoride binding to hydroxyapatite. The Patlak graphical analysis and compartmental modeling produces parametric images of less quality, but they show that the high net fluoride uptake in certain skeletal regions is due to high perfusion (K1), and that fluoride binding (k3) in these regions is within the same range as the k3 observed in other skeletal regions (Additional file 1: Fig. S2).
The group of bones with the second highest uptake were the mandible, lumbar vertebrae, ilium, and costochondral joints. As shown in Fig. 3 and in Additional file 1: Figs. S2 and S4, these uptakes were bone specific (k3). The relatively high uptake in the mandibular bone was probably due to [18F]fluoride accumulation in the dentine of teeth [18, 19]. The tibia, radius, parietal bone, and ribs showed the lowest uptake. Arvola et al. found similar order in 18F-uptake 60 min after [18F]NaF injection when measuring standardised uptake values (SUVs) from normal appearing human bones spine, pelvis, limbs, rib and skull [20].
Our results indicated that [18F]fluoride perfusion and uptake varied among the different bones. This variation must be taken into consideration when comparing, e.g., SUVs in different bone diseases in a clinical setting or when reporting the bone uptake of [18F]fluoride released from the 18F-labelled tracer. We found the highest [18F]fluoride uptake in cancellous bones, and we demonstrated that this uptake depended on perfusion. These results were consistent with a previous study by Eble et al, who revealed that, after long-term fluoride exposure, fluoride concentrations were higher in cancellous bones than in compact bones [21].
As shown in Figs. 1 and 2b (in vivo) and in Additional file 1: Fig. S3a (ex vivo), [18F]fluoride was rapidly cleared from the blood during the first few minutes after the injection. Low tracer uptake was observed in the myocardium, therefore, the blood input function for kinetic modeling of the in vivo scanned data was measured in the left cardiac ventricle. Park-Holohan et al. suggested that fluoride was rapidly transported through red blood cell membranes, and then it was available for clearance, through uptake by the bone. Therefore, they suggested that bone kinetic data for [18F]fluoride would be more accurately reported as whole-blood clearance, rather than plasma clearance [14].
Our finding that [18F]fluoride did not bind to plasma proteins was reported previously. This property is considered an advantage of [18F]fluoride, compared to [99mTc]diphosphonate, which showed 25–70% binding to plasma proteins [2]. Due to the lower molecular weight of fluoride compared to phosphonates, [18F]fluoride has a higher single-passage extraction efficiency compared to phosphonates [2]. The [18F]fluoride uptake we measured in bone marrow was consistent with the findings of Blake et al., who speculated that the uptake in red blood cells might reflect the uptake of immature erythrocytes in bone marrow [2].
The rapid, high kidney uptake demonstrated that [18F]fluoride underwent both glomerular filtration and tubular secretion. Previously, Whitford demonstrated that fluoride was rapidly cleared from the plasma, both through renal excretion and by diffusion through capillaries. That study also showed that, due to a higher glomerular filtration capacity, renal clearance of fluoride was increased at higher pH urine levels [22]. Interestingly, we found a significant difference in the amount of total excreted urine between female and male rats (Additional file 1: Fig. S3c). This might be explained by the different anatomical and hormonal features of female and male rats [23].
Organ-to-blood 18F-radioactivity ratios increased as a function of time in the eyes, testes, lung, brain, and ovaries. Some of the brain tissue uptake seen with any 18F-labelled PET tracer, with a tendency for defluorination, might actually originate from [18F]fluoride. This can be a source of error in kinetic PET analysis models of the brain.
One study limitation might have been the difference in age between female and male rats, but their average weights were the same. Moreover, all rats were juvenile, rather than adults (average ages: 3.7 months for females and 1.8 months for males). A previous study showed that the bone perfusion rate and vascular resistance did not differ between juvenile (2 months) and adult (6 months) rats [24]. Hence, we assumed that the age difference did not affect the [18F]fluoride uptake kinetics reported in this study.
Another limitation might be that, due to the limited field of view (12.7 cm) of the Inveon PET scanner, whole body scans of the rats were not possible. Hence, rats were scanned in two parts; head-to-middle and middle-to-hind legs. Furthermore, the small size of some bones (e.g., the ribs) resulted in small VOIs; hence, partial volumes and spill-over might have affected our in vivo results.