Corresponding author and reprint requests: Alan J. Fischman, Division of Nuclear Medicine, Department of Radiology, Massachusetts General Hospital, 32 Fruit Street, Boston, MA 02114, USA Tel: (617) 726-8353 Fax: (617) 726-6165
Objective: To measure tissue pharmacokinetics of trovafloxacin (CP 99,219) in normal and infected animals by both direct tissue radioactivity measurements and positron emission tomography (PET).
Method: Concentrations of Ftrovafloxacin were measured in normal and infected rats (n=6/group), at 10, 30, 60, and 120 min after injection, by radioactivity measurements. In normal rabbits (n=4) and rabbits with Escherichia coli thigh infection (n=4), tissue concentrations of drug were measured over 2 h with PET. After acquiring the final images, the rabbits were killed and tissue concentrations measured with PET were compared to the results of direct tissue radioactivity measurements.
Results: In both species, there was rapid distribution of F trovafloxacin in most peripheral organs. Peak concentrations of more than five times the MIC90 of most Enterobacteriaceae and anaerobes (>100-fold for most organisms) were achieved in all tissues and remained above this level for >2 h. Particularly high peak concentrations were achieved in the kidney (>75 μg/g), liver (>100 μg/g), blood (>40 μg/g), and lung (>10 μg/g). Even though the concentration of trovafloxacin in infected muscle was reduced (p<0.01), the peak concentration was still >4 μg/g and tissue levels remained above 2 μg/g for more than 2 h. Due to the lower concentrations that were achieved in the brain (peak ˜5 μg/g), it is expected that trovafloxacin will have limited central nervous system toxicity.
Conclusions: PET with [18F] trovafloxacin is a useful technique for non-invasive measurements of tissue pharmacokinetics.
Trovafloxacin, ((1α,5α,6α-(6-amino-3-azabicyclo [3.1.0]hex-3-yl)-1-(2,4-difluorophenyl)-6-fluoro-1,4- dihydro-4-oxo-1,8-naphthyridine-3-carboxylic acid), (CP 99,219) is a new fluoroquinolone antimicrobial agent with potent and broad-spectrum in vitro activity against both Gram-positive and Gram-negative organisms . Extensive investigations comparing the in vitro antibacterial activity of trovafloxacin with that of other fluoroquinolones have been reported [2–8]. In these studies, trovafloxacin was significantly more potent than standard fluoroquinolones against many Gram positive organisms, including Streptococcus pneumoniae, Streptococcus pyogenes, and ciprofloxacin-susceptible and -resistant staphylococci. Also, trovafloxacin had more potent activity against anaerobic organisms, while retaining potency against Enterobacteriaceae. In general, trovafloxacin had greater potency and spectrum of activity (MIC90≦ 2 μg/mL) than ciprofloxacin, temafloxacin and ofloxacin. Organisms for which MIC90S were ≧ 4 μg/mL were rare, and included Pseudomonas aeruginosa, Bacteroides fragilis, Staphylococcus haemolyticus and oxacillin-resistant Staphylococcus aureus. Trovafloxacin has also been shown to have significant in vivo activity against numerous experimental infections [9–11].
Studies of the tissue distribution of trovafloxacin in laboratory animals have demonstrated rapid absorption from the gastrointestinal tract and high concentrations in virtually all tissues (unpublished results); however, detailed tissue pharmacokinetic studies have not been reported. In humans, concentrations of drug in body fluids and a small number of tissues have been measured [12,13], but detailed studies of tissue distribution have not been done. Clearly, a precise understanding of trovafloxacin distribution in normal and infected tissues would be of value for optimizing the use of this drug in the treatment of focal and systemic infections. This information would be useful in determining dosing schedules, as well as in predicting therapeutic efficacy and drug toxicity.
Over the past several years, we have demonstrated that positron emission tomography (PET) is an extremely powerful in vivo technique for making detailed pharmacokinetic measurements in both animal models and humans [14–21]. The presence of three fluorine atoms in the native structure of trovafloxacin would make PET a useful technique for measuring pharmacokinetics if a rapid method for preparing 18F-labeled trovafloxacin could be devised. Recently, we developed a method for radiolabeling trovafloxacin by 18F-19F exchange [22,23]. This radiolabeled drug is chemically identical to the native drug, except that one or more of the fluorine atoms are 18F. Since trovafloxacin undergoes a low level of in vivo metabolism within 6–8 h after injection in humans, the time frame for most PET studies, tissue and blood radioactivity measurements should accurately reflect concentrations of intact drug (unpublished results). Thus, this reagent should be a useful radiopharmaceutical for PET studies of tissue pharmacokinetics in both laboratory animals and humans.
MATERIALS AND METHODS
Preparation of 18F-labeled trovafloxacin
18F-Labeled trovafloxacin was prepared by 18F for 19F exchange followed by HPLC purification. Briefly, 18F-prepared by the 18O(p,n)18F reaction  was dried azeotropically with acetonitrile in the presence of 4.0 mg of K2CO3 and 14.6 mg of Kryptofix 2.2.2®, 1.0 mg of trovafloxacin in 0.5 mL of dimethylsulfoxide (DMSO) was added and the mixture was heated at 160°C for 15 min. Purification was performed by reverse-phase HPLC (column: Vydac C18, 250 × 10 mm, mobile phase 83:17 phosphate buffer, pH 4.4/acetonitrile, flow rate 4 mL/min). The product was collected in a sterile glass vial, evaporated to dryness under vacuum, dissolved in sterile saline for injection and sterilized by filtration through a 0.22-μm membrane filter (Millex-GS, Millipore). This method routinely provided 18F-labeled trovafloxacin with radiochemical yields of 15–30% and radiochemical purity of >97% within 45 min. Further details of the radiolabeling procedure have been described elsewhere .
Pharmacokinetics of trovafloxacin in normal and infected rats
These studies were designed to evaluate the tissue pharmacokinetics of trovafloxacin in normal and infected rats. Groups of 24 normal and infected adult male Sprague–Dawley rats (Charles River Breeding Laboratories) weighing ˜200 g were injected via the tail vein with 18F-labeled trovafloxacin (10 mg/kg, ˜50 μCi) in 50 μL of lactate buffer, pH 4.4. At 5, 30, 60 and 120 min after injection, groups of six animals were killed and samples of blood, heart, lung, liver, spleen, stomach, intestine, kidney, muscle, testis, brain and bone were weighed and radioactivity was measured with a well-type gamma counter (LKB model no. 1282, Wallac Oy, Finland). To correct for radioactive decay and calculate the concentration of radioactivity in each organ as a fraction of the injected dose, radioactivity in aliquots of the doses was measured in parallel with the tissue samples. Assuming minimal metabolism during the time course of the study, radioactivity measurements were converted to drug concentrations by dividing by the specific activity of the injected dose. Concentrations of trovafloxacin in the brain were corrected for intravascular drug by assuming a blood volume of 4.5%. The results were expressed as micrograms of trovafloxacin per gram of tissue .
Adult male Sprague–Dawley rats (Charles River Breeding Laboratories) weighing ˜200 g were infected with Escherichia coli from a single clinical isolate. The organisms were grown on trypticase soy agar plates and individual colonies were diluted with sterile normal saline to produce a suspension containing ˜2 × 109 bacteria/mL. A 0.5-mL inoculum of the suspension was injected into the right posterior thigh muscle and the animals were allowed to recover for 24 h prior to radiopharma-ceutical injection and tissue radioactivity measurements. Previous results from our laboratory have demonstrated reproducibility of this infection model .
Pharmacokinetics of trovafloxacin in normal and infected rabbits
These experiments were designed to evaluate the tissue pharmacokinetics of trovafloxacin in normal rabbits and rabbits with deep thigh E. coli infection by PET. After acquisition of the final images, the rabbits were sacrificed and direct tissue radioactivity measurements were made. All studies were performed with co-injection of unlabeled drug.
Positron emission tomography
The rabbits were positioned prone on a plexiglas imaging table, with legs extended and separated from the pelvis. Imaging was performed with a PC-4096 PET camera (Scanditronix AB, Sweden) . The primary imaging parameters of the PC-4094 camera are: in-plane and axial resolutions of 6.0 mm FWHM, 15 contiguous slices of 6.5-mm separation with an axial field of view of 9.8 cm, and a sensitivity of ˜5000 counts/s per μCi.
All images were reconstructed with a standard filtered back-projection algorithm (in-plane resolution of 6 mm FWHM). Attenuation correction was performed with transmission data acquired with a 68Ge rotating pin source. The projection data were corrected for: non-uniformity of detector response, dead time, random coincidences and scattered radiation. Regions of interest (ROIs) were circular, with a fixed diameter of 16 mm. The lowest concentrations of radioactivity that were measured were approximately 100 nCi/mL, with a precision of ±5%. This corresponds to a quantitation limit of 0.010% injected dose/g tissue. The PET camera was cross-calibrated with a well counter by comparing the PET camera response from a uniform distribution of a 18F solution in a 20-cm cylindrical phantom with the response of the well counter to an aliquot of the same solution.
Four random-bred male New Zealand white rabbits weighing ˜3.0 kg were studied. Immediately before imaging, the rabbits were anesthetized with ketamine/rhompine and a left femoral arterial catheter was inserted for blood sampling. Each animal was placed in the gantry of the camera, with the area of the brain centered on the first slice, and [18F]-labeled trovafloxacin (10 mg/kg, ˜5.0 mCi) was injected (over 1 min) through a marginal ear vein. The total volume injected in each animal was ˜2.0 mL in lactate buffer, pH 4.4. Imaging was started immediately after injection. The rabbits were imaged by indexing the scanner table in 9.75-cm steps to yield a total of 75 slices. The duration of imaging at each bed position was 1.0 min. Image sets were acquired starting at 0, 5, 15, 30, 60 and 120 min after injection, and arterial blood samples were collected at 1, 5, 10, 30, 60, 90 min and just prior to killing. After acquisition of the final image, the animals were killed with a lethal dose of sodium pentobarbital. Samples of heart, lung, liver, spleen, kidney, prostate, brain, bone, bladder, adrenal, pituitary, prostate, stomach and intestine were collected and weighed, and radioactivity was measured with a well-type gamma counter. To correct for radioactive decay and permit calculation of radioactivity in each organ as a fraction of the injected dose, aliquots of the doses were counted in parallel. Concentrations of trovafloxacin in the brain were corrected for intravascular drug by assuming a blood volume of 4.5%. The results were expressed as micrograms trovafloxacin per gram of tissue.
Four male New Zealand White rabbits weighing ˜3.0 kg were infected with E. coli (1 × 1010 organisms) as described above. The animals were allowed to recover for 24 h prior to imaging.
The procedures for anesthesia, catheter placement, drug administration and blood sampling were identical to the methods used for studying normal rabbits. However, the imaging procedure was modified as follows. At 5, 10, 15, 20, 25, 30, 60 and 120 min after injection, 5-min images of the thigh muscles were acquired. Between 30 and 60 min and 65 and 120 min after injection, whole body images were acquired as described above. After acquisition of the final images, the rabbits were killed and biodistribution was measured.
The data were evaluated statistically by analysis of variance (ANOVA) followed by Duncan's new multiple range test . For the biodistribution experiments in rats, two-way ANOVA with a linear model in which time after injection and infection status were the classification variables, was employed for each tissue: [Trovafloxacin] = Time + Infection + (Time X Organ). For the rabbit experiments, two-way ANOVA with a linear model in which infection status and organ were the classification variables was used: [Trovafloxacin] = Infection + Organ + (Infection X Organ). Drug concentrations in normal and infected muscle were compared by one-way ANOVA. In order to describe the blood clearance of trovafloxacin in terms of a limited number of parameters, the time dependence of the blood concentration of drug was fitted to a biexponential function by unweighted non-linear least squares using the program PROC NLIN (SAS Institute). All results are expressed as mean ± SEM.
Pharmacokinetics of trovafloxacin in uninfected tissues of normal and infected rats
Figure 1 shows the concentration of trovafloxacin (μg/g) in uninfected tissues of normal and infected rats at 5, 30, 60 and 120 min after injection. With the exception of brain, relatively high concentrations of drug were detected in all tissues. For most tissues, the highest concentration of drug was measured at 5 min after injection and decreased at later times. In stomach and testis accumulation was delayed, while in intestine accumulation increased continuously over the time course of the study.
For all tissues, ANOVA demonstrated significant main effects of time after injection; stomach p<0.05, all others p<0.00001. Significant main effects of infection were observed for: blood p<0.005, kidney p<0.00001, intestine p<0.05, testis p<0.01 and brain p<0.01. Significant infection–time interactions were not detected for any tissue. Based on the assumption that 100% of the drug is contained within the vasculature immediately after injection, within 10 min after injection > 90% of the radiolabeled drug was cleared from blood into the tissues.
Trovafloxacin concentrations in infected rats were lower in: blood (5 and 30 min, p<0.01); spleen (30 min, p<0.01); adrenal (30 min, p<0.01); intestine (30 min, p<0.05; 60 min, p<0.01); and bone (5 min, p<0.01). In contrast, trovafloxacin concentrations in infected animals were higher in: heart (60 min, p<0.05); liver (60 min, p<0.01); kidney (5 and 60 min, p<0.01; 120 min, p<0.005), testis (60 min, p<0.01; 120 min, p<0.05); and brain (120 min, p<0.01). In stomach, higher concentrations were measured in infected rats at 60 min p<0.05) and in normal rats at 5 min p<0.05).
Peak concentrations ranged from >35 μg/g for intestine, liver and kidney to >10 μg/g in lung, spleen, adrenal, stomach, heart, muscle, blood and bone, to >9 μg/g in the spleen and heart, to > 5 μg/g in the testis, to < 0.5 μg/g in the brain. Even at 2 h after drug administration, concentrations of trovafloxacin were > 2 μg/g in most tissues.
Pharmacokinetics of trovafloxacin in infected muscle of rats
Figure 2 shows the concentration (μg/g) of trovafloxacin in thigh muscle of normal and infected rats at 5, 30, 60 and 120 min after injection. In both normal and infected muscle, the highest concentration of drug was measured at 5 min after injection and decreased thereafter. ANOVA demonstrated significant main effects of time (p<0.0001), infection (p<0.01) and infection–time interaction (p<0.00001). At 5 min, the trovafloxacin concentration was greatest in uninfected muscle of infected animals (p<0.01), and the concentration in infected muscle was greater than in muscle of uninfected animals (p<0.01). At 60 and 120 min, the concentration of trovafloxacin in infected muscle was higher than in uninfected muscle of infected animals (p<0.01 and p<0.05) but not significantly different from that in normal animals. At 1 and 2 h after administration, drug concentrations in infected muscle were > 4 μg/g and > 2 μg/g.
PET imaging and biodistribution of trovafloxacin in rabbits
Figure 3 shows representative PET images of a rabbit 30 min after injection of [18F]trovafloxacin with 10 mg/kg of unlabeled drug. From these images, it is apparent that relatively high concentrations of trovafloxacin are achieved in all tissues, with the exception of brain. In most tissues, high levels of accumulation were achieved within 20 min after injection and decreased over the next 100 min. Figure 4 shows that blood decline of trovafloxacin in normal and infected rabbits is well described by biexponential functions (fitting parameters expressed as value ± SD):
Although infection did not have a significant effect on the rate constants for the α or β phases of trovafloxacin clearance (p<0.1), the contribution of the α phase was decreased and the contribution of the β phase was increased (p<0.01). The apparent volumes of distribution (Vd) of drug in normal and infected animals were 700.2±15.7 and 724.9±38.9 mL (p=NS).
Figure 5 illustrates tissue accumulation curves for trovafloxacin in normal (n=4) and infected (n=4) rabbits measured by PET. Since there was no significant difference in drug concentrations between uninfected tissues of normal and infected animals, these data were pooled. Peak concentrations of trovafloxacin were: >75 μg/g in kidney (not shown); >100 μg/g in liver; >40 μg/g in blood; >10 μg/g in lung; >5 μg/g in brain; > 8 μg/g in uninfected muscle. The concentration of trovafloxacin in infected muscle was significantly (p<0.01) lower than in normal muscle at all imaging times; however, peak concentration was still >4 μg/g and remained at >2 μg/g at 2 h after injection. Figure 6 shows the biodistribution of trovafloxacin in uninfected tissues of normal and infected rabbits at 2 h after injection. ANOVA demonstrated significant main effects of tissue (p<0.0001) and infection (p<0.005). However, tissue–infection interaction was not significant (p<0.4) and differences in drug concentration were not detected in any specific tissue. Even at 2 h post-administration, concentrations of > 2 μg/g were present in most tissues (>8 μg/g in kidney, liver, gastrointestinal tract, lung, prostate, stomach, bladder and spleen; brain concentrations were >0.4 μg/g). Figure 7 shows the concentrations of trovafloxacin in normal and infected rabbit muscle. Although drug concentration was lower in infected muscle, in contrast to the imaging results, ANOVA failed to demonstrate a significant effect. In general, trovafloxacin concentrations determined by direct measurements of radioactivity in tissue samples were in excellent agreement with the PET results.
The pattern of distribution of trovafloxacin in normal and infected rats determined with [18F]trovafloxacin is generally consistent with the results of whole body autoradiographic studies at one time point after injection of 14C-labeled drug (unpublished results). Also, the pattern of distribution of trovafloxacin in rabbits measured by both PET and direct radioactivity measurements was similar to that seen with rats. The monotonic decrease in trovafloxacin concentration in kidney and liver is most probably related to the fact that these are clearance organs. Similarly, the rapidly decreasing concentration of drug in lung and muscle may be due to rapid equilibration in these organs. The low level of trovafloxacin accumulation in bone is consistent with its proposed mechanism of metabolism via conjugation with minimal defluorination [28,29]. This observation supports the utility of [18F]trovafloxacin prepared by 18F–19F exchange for pharmacokinetic studies . In rats, at 5 min after injection, trovafloxacin concentration in infected muscle was greater than in muscle of uninfected animals but lower than in uninfected muscle of infected animals. At 30 min, similar concentrations were measured in all three tissues, while at 60 and 120 min, concentrations in infected muscle and muscle of uninfected rats were similar to each other but greater than concentrations in uninfected muscle of infected animals. In contrast, drug concentrations in uninfected muscle of infected rabbits were greater than in infected muscle or muscle of uninfected animals. Currently, we have no explanation for these differences between rats and rabbits.
Since the major purpose of this study was to develop and validate a PET method that could be applied to humans, it is fortunate that the primary route of trovafloxacin metabolism in humans is conjugation, which does not affect the 18F labels. For human studies, the spatial resolution and quantitative nature of PET measurements will allow the evaluation of drug distribution in small volumes of tissue. Our previous experience with 18F-labeled antimicrobial agents has demonstrated the validity of this approach [14–21]. Based on the current biodistribution data, preliminary Medical Internal Radiation Dose (MIRdose) calculations  indicate that ˜20 mCi of [18F]trovafloxacin can be administered to humans without delivering a radiation burden in excess of 20 mGy to any organ (unpublished results). With this dose of radioactivity, pharmacokinetics can be studied over 8 to 10 h by PET.
The use of PET to study pharmacokinetics has many advantages over more conventional techniques. Although changes in drug concentrations in tissues could be determined by radioactivity measurements of excised tissues or quantitative autoradiography using many animals, PET allows multiple measurements in the same subject at different times and in a variety of physiologic or pathologic states. A major benefit of PET is the quantitative nature of the measurement. Even more importantly, PET is the only non-invasive technique that can be used to measure tissue pharmacokinetics quantitatively in humans.
The concentrations of trovafloxacin achieved in normal and infected tissues of rats and rabbits appear promising when compared to the sensitivity of a wide range of microorganisms to trovafloxacin. Thus while the MIC90S for virtually all the Enterobacteriaceae and anaerobes are <0.5 μg/mL [2,4–8,31], peak concentrations of drug are well over 5 μg/g in all extracranial tissues, with sustained concentrations above the MIC90S in virtually all tissues for at least 2 h. In particular, the animal studies performed thus far suggest that this drug should be especially useful in the treatment of gastrointestinal, urinary tract, hepatobiliary and pulmonary infections, and should have promise in the treatment of infections of many other sites. Even though the concentration of trovafloxacin in infected muscle was reduced (p<0.01), the peak concentration was still >4 μg/g and tissue levels remained above 2 μg/g for more than 2 h. The low concentrations of drug achieved in the brain indicate that central nervous system toxicity, an important issue with fluoroquinolones, may not be a significant problem. However, since these concentrations are higher than the MIC90 for S. pneumoniae (˜ 0.2 μg/mL), it is possible that trovafloxacin may be effective for treating central nervous system infections caused by this organism [32,33].
In summary, we have developed a method for preparing [18F]trovafloxacin and demonstrated its applicability for the non-invasive measurement of tissue pharmacokinetics by PET. The studies presented here demonstrate that there is rapid accumulation of drug in most tissues and the concentrations achieved are well above the MIC90S of most relevant pathogens. These results indicate that future PET studies in human subjects will yield valuable insights into the clinical application of this important new drug.
This work was supported in part by Pfizer Central Research, Groton CT.