Pharmacokinetic/pharmacodynamic (PK/PD) evaluation of tulathromycin against Haemophilus parasuis in an experimental neutropenic guinea pig model

The objective of the study was to develop an ex-vivo PK/PD model of intramuscular (IM) administration of tulathromycin and to test its efficacy against Haemophilus parasuis (H. parasuis) infection in intraperitoneal-inoculated neutropenic guinea pigs. The pharmacokinetics (PKs) of tulathromycin at doses of 1 and 10 mg/kg in H. parasuis-infected neutropenic guinea pig were studied by high-performance liquid chromatography-tandem mass spectrometry (HPLC-MS/MS). In vitro minimum inhibitory concentration (MIC), minimum bactericidal concentration (MBC), mutant prevention concentration (MPC), post-antibiotic effect (PAE) and dynamic time-kill curve experiments were carried out using H. parasuis strain 13R. Tulathromycin exhibited concentration-dependent activity and PAE persisted long after administration of the antibiotic. The ratio of the 24-h area under the concentration–time curve (AUC) to MIC in serum (AUC24h/MICserum) was recognized as an important PK/PD parameter that positively correlated with the in vitro antibacterial effectiveness of tulathromycin (R2 = 0.9961 or R2 = 1). For the 1 and 10 mg/kg treatments with tulathromycin, the values of AUC24h/MIC for H. parasuis bacteriostatic action, bactericidal action and virtual bacterial eradication were respectively 22.73, 34.5 and 88.03 h for the 1 mg/kg treatment and respectively 24.94, 30.94 and 49.92 h for the 10 mg/kg treatment. In addition, we demonstrated that doses of 7.2–8.0 mg/kg of tulathromycin resulted in high eradication rates (99.99%). Using a previously published conversion factor of 0.296, we were able to estimate an approximate dose, 2.1–2.4 mg/kg, that should also obtain high eradication rates in the target animal, pigs. This study can help optimize tulathromycin efficacy against H. parasuis infections in swine farming.


Introduction
Tulathromycin is a semi-synthetic macrolide approved solely for veterinary use by the US (Mindray BC-2800Vet, Shenzhen, China). An H. parasuis standard strain of serotype 13 (13R) was used to infect the neutropenic guinea pigs via intraperitoneal injection, specifically, a single inoculum of a 0.2 mL aliquot solution containing approximately 10 9 CFU of the H. parasuis strain, which was a 95% infective dose (ID 95 ) that was previously determined in pilot studies. Ten guinea pigs were used to detect blood and lung tissue bacterial loads in a 72 h-period, and 256 guinea pigs were used for PKs of tulathromycin in a 168 h-period. To obtain antimicrobial PK data, infected neutropenic guinea pigs were given single IM doses at two levels, 1 or 10 mg/ kg, via tulathromycin Injection (Draxxin1) (Zoetis, New York, USA).

Sample collection
To quantify pathogen load in the H. parasuis intraperitoneal infection model, blood samples were collected from the heart before euthanasia and viable counts (CFU/mL) were detected immediately, and lung tissues were removed at sacrifice by using an overdose of sodium pentobarbital on each animal at about 2 h, when typical symptoms of the respiratory tract were observed, which included appearing depressed, crowding together, closing eyes, drooping, coughing, and abdominal breathing, and 72 post-challenge. No guinea pigs died in a 72hperiod. Viable counts (CFU/g) of bacteria in lung tissues were detected within one hours of tissue removal. Five guinea pigs were sampled at each time point.

Analysis method
Chemicals and reagents. Tulathromycin (99.85%) was supplied by Shandong Lukang Pharmaceutical Co., Ltd (Shandong, China). Roxithromycin was purchased from National Institutes for Food and Drug Control (Beijing, China) and used as an internal standard. HPLC grade methanol, acetonitrile and formic acid were used for drug analysis. Ultrapure water was provided by a Milli-Q ultrapure water purification system (Milli-Q Millipore Corp.).
Viable count. Lung tissue (0.5 g) was added into 0.5 mL of phosphate-buffered saline (PBS), and centrifuged at 7200 rpm for 15 seconds by a Precellys Evolution Super Homogenizer (Bertin Technologies, France). Aliquots of 0.1 mL supernatant were used in 10-fold serial dilutions to obtain viable count data (CFU/mL). We removed 20 μL from each dilution to spread on tryptic soy agar (TSA) plates containing 10 μg/mL nicotinamide adenine dinucleotide (NAD, Sigma, Inc., USA) and 5% bovine serum (Gibco1, New Zealand). The colonies were counted (CFU) after incubation for 36-48 h at 37˚C. Blood samples were collected from the heart before euthanasia and viable count data was collected. The lowest detectable count was 10 CFU/mL. All samples were performed in triplicate.
Drug analysis. Serum was obtained by centrifuging blood samples at 4,000 rpm for 10 min and immediately stored at -20˚C until analysis (within one month of sampling). Tulathromycin concentrations in serum and lung tissue were determined via HPLC-MS/MS method as described previously [15,16]. The free fraction (fu) of tulathromycin in guinea pig serum was determined in triplicate using Amicon Centrifree Micropartition devices (Millipore, Bedford, MA, USA) with a 10000 Nominal Molecular Weight Limit according to the manufacturer's instructions. To measure PK parameters of guinea pig serum, 0.5 mL was added into an ultrafiltration device and then centrifuged at 12000 g for 45 min at 25˚C. Measurements for PK parameters were taken using samples obtained at 0.5, 2, 8, 24 and 48 h. The concentration of tulathromycin was determined by HPLC-MS/MS. Pharmacokinetics analysis. The PK parameters of tulathromycin in serum were analyzed by WinNonlin software (version 5.2; Pharsight, CA, USA). A WinNonlin model 200 was used for non-compartmental analysis of the concentration-time data. PK parameters were expressed as mean values ± standard deviations (SD).
In vitro and ex vivo susceptibility studies and PAE. H. parasuis was cultured in as a previous report [17]. A total of 94 strains of H. parasuis strains isolated from swine between 2014 and 2016 were used in this study. The MIC and MBC of tulathromycin against H. parasuis were determined in both CAMHB cultures and serum by the micro dilution method according to protocols acquired from the Clinical and Laboratory Standards Institute [18]. The concentrations ranged from 0.015 to 8 μg/mL. Tests were conducted in triplicate and included growth controls (H. parasuis in the media only), and germ-free controls (blank media only). The MIC value is defined as the lowest tulathromycin concentration exhibiting no visible growth of H. parasuis when bacteria were cultured at 37˚C for 24 h. The MBC is defined as the lowest drug concentration which resulted in a 99.9% reduction of bacterial density when bacteria were cultured at 37˚C for 24 h.
The measure of MPC of tulathromycin against H. parasuis 13R was determined by the agar method [19]. A final concentration of *3 × 10 10 CFU/mL was used in determining MPC. Samples (100 μL) were plated onto TSA agar containing various concentrations of tulathromycin obtained by successive two-fold dilutions. The MPC was measured at MIC levels of 1, 2, 4, 8, 16, 32 and 64. All plates were incubated at 37˚C for 72 h and then examined for growth. The MPC represents the lowest antibiotic concentration where no bacterial growth was observed while under anaerobic conditions. Samples were performed in triplicate.
The PAE of tulathromycin was estimated by the removal of drug method [20]. H. parasuis 13R was incubated with 1, 2 and 4 MIC of tulathromycin. After incubating for 1 or 2 h, he drug was removed by dilution 1000 times with fresh medium. The viable counts of H. parasuis were determined at 1, 2, 4, 6, 8, 10 and 12 h. The PAE was calculated as follows [20]: PAE = T-C;, where T and C represent the time periods required for viable counts of bacteria to increase by 1-log10 CFU in the drug removal phase for the treatment and the untreated control groups, respectively.
In vitro and ex vivo time-killing curves. In vitro time-killing curves of tulathromycin against H. parasuis were obtained after determination of MIC and MBC values. Serial concentrations of tulathromycin were prepared in CAMHB and guinea pig serum ranging from 0.125 to 32 MIC before bacterial inoculation (10 6 CFU/mL). Growth and sterile controls were performed at the same time. The total volume for each concentration was 4 mL. The cultures were incubated at 37˚C for 24 h, and the viable counts of bacteria were determined at 0, 2, 4, 6, 8, 10, 12 and 24 h of incubation time. The limit of detection was 10 CFU/mL. All samples were performed in triplicate.
All serum samples obtained from guinea pigs (0-168 h) after IM injection of tulathromycin were used to establish ex-vivo time-killing curves. A control serum was prepared from samples collected from the same guinea pig before administration. All serum samples were pre-filtered through a 0.22 μm membrane to eliminate any pathogens. In order to reduce potential variability in growth rates of bacteria between the in vitro and ex-vivo experiments, the same volume of 4 mL as CAMHB samples and same incubation time of 24 h was used as described above. We averaged the values measured from the sera of each group of eight guinea pigs per time point to plot ex vivo time-killing curves. The viable counts of bacteria were determined as described above.
PK/PD modeling and data analysis. The surrogate markers of antibacterial effectiveness of tulathromycin, AUC 168h /MIC, T>MIC, and C max /MIC, were calculated using in vitro MIC values and PK parameters obtained from IM administrations of tulathromycin in serum samples. The effectiveness of tulathromycin was expressed as changes in log 10 CFU after 24 h of incubation of serum samples used to establish ex-vivo time-killing curves. The in vitro PK/ PD relationship of tulathromycin was described using a sigmoid inhibitory E max model with the WinNonlin software (Version 5.2; Pharsight, CA, USA) and the equation is as follows: where E is the antibacterial effect measured as the reduction in log 10 CFU/mL after administration of tulathromycin compared to the log 10 CFU/mL in the untreated control group; E max is the reduction in log 10 CFU/mL for the untreated control guinea pigs; E 0 is the maximum reduction after administration and represents the maximum antibacterial effect; Ce is the AUC 0-24h /MIC parameter; EC 50 is the AUC 0-24h /MIC serum value required to achieve 50% of the maximal antibacterial effect; and N is the Hill coefficient that describes the steepness of the AUC 0-24h /MIC serum and effect curve. Dose regimen prediction. In order to determine an optimal dose regimen to apply for treatment of H. Parasuis in veterinary medicine, the dose required for a given level of antibacterial activity is predicted by the equation [21]: where Dose is the optimal dose to produce and sustain antibacterial activity for 10 days (mg/ kg/10-day); Cl is the serum clearance; factor is the dimensionless numerical value of AUC/ MIC serum ; MIC 90 is the 90th percentile of the MIC distribution; fu is the free drug fraction; and F is the bioavailability of tulathromycin.

Pharmacokinetics of tulathromycin
Animals were severely granulocytopenic, and clinical symptoms and bacteriological examinations verified that neutropenic guinea pigs were successfully infected with H. parasuis (Table 1, Fig 1). The PK parameters of the two dose levels are shown in Table 2 and the serum concentration-time profiles are illustrated in Fig 2. After IM administration, tulathromycin was rapidly absorbed and its concentrations peaked at 0.5 h at mean values (C max ) of 1079.5 and 3486.25 ng/mL for the 1 mg/kg and 10 mg/kg doses, respectively. The mean elimination half-lives of the two respective dose levels were 25.8 and 26.9 h, which indicate slow elimination rates. The mean area under the concentration-time curve (AUC 0−168h ) were 11019.6 and 57182.5 ng.h/ mL, respectively. Mean body clearance rates (Cl) were 88.1 and 172.0 mL/kg/h, respectively. Additionally, the free fraction (fu) of tulathromycin in guinea pig serum was 0.56-0.74.

In vitro and ex vivo antimicrobial activities of tulathromycin
In vitro time-kill curves obtained for antimicrobial activity against H. parasuis 13R for different concentrations of tulathromycin in the range of 0.125-32x of the MIC are shown in Fig 3. The concentrations less than the MIC of tulathromycin exhibited similar levels of antimicrobial activity against H. parasuis. However, when tulathromycin concentrations were higher than the MIC, the bacteriocidal activity gradually improved due to the increase in drug concentration, which supports the notion that the concentration-dependent response is a characteristic of tulathromycin activity. The time-kill curves of tulathromycin against H. parasuis 13R in the range of 0.125-32x of the MIC were also obtained from blank guinea pig serum and infected guinea pig serum samples, both of which were treated with tulathromycin (Fig 4). Similar antimicrobial activity of tulathromycin against H. parasuis was observed in guinea pig serum. The bactericidal activity increased with increasing concentration of tulathromycin up to 2 MIC. Additionally, when bacterial counts were reduced by roughly 4-log 10 CFU/mL with increasing concentration of tulathromycin, it occurred at a shorter time period of 2-4 h. Data from guinea pig serum samples that had IM administrations with one of the two dose levels of tulathromycin and their blood collected at different time points (0-168 h) were used to determine an ex-vivo killing rate. In the 1 mg/kg treatment (Fig 5), no bactericidal activity was observed for serum collected at 144 or 168 h. However, in the 10 mg/kg treatment and within the range of concentrations of 0.31 to 3.45 μg/mL, the number of bacteria decreased slightly after 0.083-48 h (Fig 6). Additionally, no bactericidal activity was measured in serum collected at 168 h.

PK/PD integration and dose regimen prediction
The ratios of PK/PD parameters C max /MIC serum , AUC 168h /MIC serum , C max /MBC serum , AUC 168h /MBC serum , C max /MPC CAMHB , AUC 168h /MPC CAMHB of tulathromycin activity against H. parasuis are shown in Table 5. According to these ratios and the sigmoid E max model, the PK/PD indices AUC 0-24h /MIC serum and C max /MIC were the best predictors of tulathromycin efficacy against H. parasuis. AUC 0-24h /MIC, %T>MIC, C max /MIC and the degree of antibacterial effect are displayed in Fig 7 and Fig 8. We quantified PK/PD parameters and the antibacterial effects of tulathromycin and presented them in Table 6. We characterized tulathromycin efficacy by three levels: (1)  After testing two dose levels, 1 mg/kg and 10 mg/kg, by a single intravenous administration in infected guinea pigs, we were able to predict the dose of tulathromycin necessary to virtually eradicate an H. parasuis infection in guinea pig by using the values of AUC 24h /MIC from the PK/PD model and of fu and MIC 90serum obtained in our study. We estimated that for the 1 mg/kg and 10 mg/kg treatments, a dose range between 7.2 and 8.0 mg/kg would achieve the virtual eradication of H. parasuis over a ten-day period.

Discussion
H. parasuis can cause systemic infection in pigs, including anorexia, depression tremors, posterior paresis or lateral recumbency, and incoordination [22][23][24]. In the present study, we established a neutropenic guinea pig model where we successfully infected with H. parasuis by intraperitoneal injection, to study the efficacy of tulathromycin and to eliminate the influence of individual differences due to immunity among guinea pigs. Evaluation of the H. parasuis infection model relied first on clinical symptoms, and then on bacteriological assays. In the present study, clinical symptoms and bacteriological examinations were observed and detected in infected animals from 2 h post-infection, which indicated guinea pigs infected with H. parasuis produce clinical and dissection symptoms similar to infected pigs, and the guinea pig model can replace the target animal to study H. parasuis.
In this study, after single IM doses of 1 or 10 mg/kg of tulathromycin, the mean T 1/2 were 25.8 h and 26.9 h, respectively. These results are similar to results of previously published articles on laboratory animals other than guinea pigs such as mice [25] and rabbit [26]. However, our mean T 1/2 were shorter than the results of studies that used target animals, such as pig, cattle, goats, and horses [27][28][29][30][31][32]. Other studies' PK parameters were similar to our pharmacokinetic results, such as the rapid rate of absorption, wide extent in systemic availability, and extensive tissue distribution [25][26][27][28][29]31]. Additionally, the free fraction (fu) of tulathromycin in guinea pig serum (0.56-0.74) was similar in range to a previous report (0.53-0.68) using swine serum [13]. Between 2000 and 2002, studies in the U.S. and Canada reported MICs of tulathromycin against H. parasuis from 0.25 to >64 μg/mL and an average MIC 90 of 2 μg/mL [13], and between 2008 and 2011, studies in the Czech Republic reported MICs from 0.5 to 64 μg/mL and an average MIC 90 of 8 μg/mL [33]. In the present study, the MIC of tulathromycin against H. parasuis was from 0.06 to 8 μg/mL for 94 bacterial strains and MIC 90 was 0.5 μg/mL in the CAMHB, whereas in serum, the MIC was 0.0075 to 0.25 μg/mL and the MIC 90 was 0.06 μg/ mL. The CAMHB/serum ratio of MIC ranged from 8.33-33.33, which indicated that the MIC measured in serum was significantly lower than the MIC measured in broth. Similar results have been reported in several previous studies [34][35][36]. The MBC measured in this study showed similar CAMHB/serum ratios to the MIC. This finding and findings of other researchers indicate that tulathromycin may have a strong effect on serum, and thus, the MIC in serum is likely more applicable than in CAMHB to establish a rational dosing regimen in in vivo or in vitro antimicrobial activity studies [34][35][36][37][38].  Ex-vivo PK/PD of tulathromycin has been studied recently, against Pasteurella multocida [35] and Streptocossus suis [34] in pigs, and against Mannheimia haemolytica and Pasteurella multocida [36] in cattle. To the best our knowledge, most macrolide drugs are time-dependent antibacterial agents, where the efficacy of these drugs depends on the time period at which the drug concentration is above the MIC. For time-dependent antibacterial agents, T>MIC and AUC/MIC were the best indices to use in PK/PD modeling [5,39]. Recent studies have reported different bactericidal activity of tulathromycin against different bacteria, for example, a time-dependent killing pattern against Streptococcus suis [34] and an in vitro concentrationdependent bactericidal activity against Haemophilus somnus [40,41] and Pasteurella multocida [35]. In the present study, the evaluation of activity of tulathromycin in vitro showed that tulathromycin had concentration-dependent bactericidal activity against H. parasuis, which supports a previous report that also concluded a positive relationship between drug concentration and bacterial killing [5].
For concentration-dependent agents like tulathromycin, either of the two ratios C max /MIC or AUC/MIC, out of all the PK/PD indices that have been researched, have shown an association with clinical outcomes [5]. In the present study, the PK/PD parameters that were better indicators of the efficacy of tulathromycin were C max /MIC and AUC/MIC. The concentrations of tulathromycin in serum at different time intervals empirically support that tulathromycin's effect on H. parasuis is concentration-dependent, thus indicating that C max is a suitable parameter to evaluate the PKs of a drug in killing bacteria [5]. Bactericidal activity significantly increased when concentration of tulathromycin added in blank serum increased as indicated by approximately up to 2 times the MIC in serum or 4 times the MIC in CAMHB in this study, which showed a high eradication rate of the infection. These results suggest that the concentration of tulathromycin when above the MIC is important for tulathromycin efficacy. However, because we observed that PAE of tulathromycin increased with in vitro time of exposure and concentration of the antibiotic, evaluation of tulathromycin efficacy by C max /MIC would be affected. Therefore, we, as well as other researchers, conclude that AUC/MIC is the better predictor of tulathromycin efficacy [34][35][36].  Table 5. Ratios of PK/PD data sampled from guinea pigs after IM administration of tulathromycin at 1 or 10 mg/kg (Mean ± SD).

Parameters
Unit Values The measure of MPC represents the lowest drug concentration that can prevent the growth of first-step resistant mutants, and is the highest concentration within the mutant selection window (MSW), whose lowest concentration is the MIC. The ratio of AUC 0-24 /MPC is known to be an indicator of selection of antimicrobial resistance [42][43][44][45][46]. An an in vitro study of marbofloxacin determined that an AUC 0-24 /MPC > 25 h could prevent resistance selection in pigs with H. parasuis infection [42]. In the present study, the MPC against H. parasuis (13R) was 2.048 μg/mL (in vitro), four times higher than the MIC (0.5μg/mL, in vitro), and the AUC 0-24h / MPC CAMHB was 5.4 ± 1.2 h. Therefore, the use of AUC/MPC may be an advantage in maximizing efficacy and reducing the incidence of drug resistance than the use of AUC/MIC. Although we knew that tulathromycin can have a strong effect on serum, however, because of we did not obtain the MPC in serum, thus, AUC/MPC was not used for the prediction of dose regimen in this study.
Integration of PK/PD for dose optimization of tulathromycin treatment via an in vitro model has been reported by some researchers for target animals. In the present study, 7.2 or 8.0 mg/kg were our dose estimates to achieve the virtual eradication of H. parasuis over a tenday period in guinea pig. Furthermore, we applied a dose conversion coefficient from guinea pig to pig [47] of 0.296 to our estimates to obtain equivalent doses of 2.1 mg/kg or 2.4 mg/kg for the target animal, pig. Our result is lower than the previous studies, 13.25 mg/kg against Pasteurella multocida [35] and 3.56 mg/kg against Streptococcus suis [34], which were higher than the recommended dose (2.5 mg/kg). In contrast, results in this study is similar to report of an estimated dosage of 2.52 mg/kg of tulathromycin against Pasteurella multocida [36]. The difference may be explained by the variable responses of different bacteria to tulathromycin which likely results in different PK/PD models. Our study supports the current marketed dose of tulathromycin as an appropriate dose and that the concentration of tulathromycin achieved in serum with this dosage is biologically and clinically relevant [36].
The main limitation in this study was that we did not take into account that tulathromycin specifically targets lung tissue [25,27,29]. In addition, differences between in vivo and in vitro conditions were not considered, for example, in in vitro conditions, the concentration of antibacterial decline was invariable, and because bacteria is continuously exposed to fixed concentration of antibiotics for a defined period of time, this ignores the natural rates of body clearance in drug metabolism of the animal [48]. Thus, the estimated dose regimen should be validated in swine in a future study.

Conclusions
The present study characterized the in vitro efficacy of tulathromycin against H. parasuis in a neutropenic infected guinea pig model. The in vitro data showed that tulathromycin exerted concentration-dependent bactericidal activity against H. parasuis, had a greater effect in serum than in CAMHB, and had long PAE. The data from this study can help improve tulathromycin efficacy with respect to bacteriological and clinical outcomes by providing a rational approach to the design of optimal dosage regimens (2.1-2.4 mg/kg) for target animals.

Author Contributions
Data curation: Yongda Zhao.  � E 0 is the change in log 10 CFU/mL in the control sample (absence of tulathromycin); E max is the difference in effect between the greatest amount of growth (observed in the growth control, E 0 ) and the greatest amount of mortality.
EC 50 is the AUC 0-24h /MIC serum value at which a 50% reduction in bacterial counts was observed. N is the Hill coefficient that describes the steepness of the AUC 0-24h /MIC-effect curve. https://doi.org/10.1371/journal.pone.0209177.t006