In vitro Dynamic Pharmacokinetic/Pharamcodynamic (PK/PD) study and COPD of Marbofloxacin against Haemophilus parasuis

Background Haemophilus parasuis (H. parasuis) can invade the body and cause systemic infection under stress conditions. Marbofloxacin has been recommended for the treatment of swine infections. However, few studies have investigated the PK/PD characteristics and PK/PD cutoff (COPD) of this drug against H. parasuis. Results MICs of marbofloxacin against 198 H. parasuis isolates were determined. The MIC50 and MIC90 were 2 and 8 mg/L, respectively. An in vitro dynamic PK/PD model was established to study the PK/PD relationship of marbofloxacin against H. parasuis. The PK/PD surrogate markers Cmax/MIC, Cmax/MPC (the maximum concentration divided by MIC or mutant prevention concentration (MPC)) and AUC24h/MIC, AUC24h/MPC (the area under the curve during the first 24 h divided by MIC or MPC) simulated the antimicrobial effect of marbofloxacin successfully with the R2 of 0.9928 and 0.9911, respectively. The target values of 3-log10-unit and 4-log10-unit reduction for AUC24h/MPC were 33 and 42, while the same efficacy for AUC24h/MIC were 88 and 110. The COPD deduced from Monte Carlo simulation (MCS) for marbofloxacin against H. parasuis was 0.5 mg/L. The recommended dose of marbofloxacin against H. parasuis with MIC ≤ 2 mg/L was 16 mg/kg body weight (BW). Conclusions The PK/PD surrogate markers AUC24h/MIC, Cmax/MIC and AUC24h/MPC, Cmax/MPC properly described the effects of marbofloxacin. Marbofloxacin can achieve the best efficacy at dosage of 16 mg/kg BW for strains with MIC values ≤ 2 mg/L, therefore, it is obligatory to know the sensitivity of the pathogen and to treat animals as early as possible. The very first COPD provide fundamental data for marbofloxacin breakpoint determination.

most important factor for the emergence and dissemination of resistance is the exposure, especially exposure to sub-optimal drug concentrations [9]. The PK/PD modelling which helps determine exposure-response relationships is of great importance in determining antimicrobial regimens administered to animals to attain appropriate effects [10]. The Fluoroquinolones and Cephalosporin of 3th and 4th generations that have been re-licensed in Europe took into account not only the classical paradigm of concentration-dependent dosage but also the PK/PD indices best descried the effects and minimized the emergence of resistances, such as AUC 24h /MIC, C max /MIC, the percent time that drug concentrations were above the minimum inhibitory concentration (%T > MIC), AUC 24h /MPC, the percent time that marbofloxacin concentrations were above the mutant prevention concentration (%T > MPC) or the mutant selection window (TMSW). Marbofloxacin, a third generation fluoroquinolone, has been developed solely for veterinary treatment. It acts as a concentrationdependent bactericidal agent against Gram-negative and gram-positive bacteria [11]. Marbofloxacin has been recommended by the European Committee and China for the treatment of swine infections with a dosage regimen of 2 mg/kg/24 h BW for three to five days. However, the PK/PD relationship of it against H. parasuis is sparse. Susceptibility breakpoint for an antimicrobial may assist in determining whether an antibacterial is potentially useful in the treatment of a bacterial infection. Knowing whether an antimicrobial is useful will promote prudent use of antimicrobial drugs. Breakpoints should be set prior to an antibacterial being used clinically or at the time of an approved use. Its setting requires integration of knowledge of the wild-type distribution of MICs, the PK/ PD relationship of an antibacterial, and clinical outcomes of infections when the antibacterial is used [12]. Veterinary susceptibility breakpoints are developed by the Clinical and Laboratory Standards Institute (CLSI) subcommittee on Veterinary Antimicrobial Susceptibility Testing (VAST) [13]. At this time, however, no veterinary specific clinical breakpoints of marbofloxacin have been established for swine disease caused by H. parasuis.
CO PD determined by MCS that considers pharmacokinetic variation in target animals and PK/PD indices assisted in the defining of susceptibility breakpoints from the perspective of exposure-response relationship [14]. This method has also been used by regulatory agencies such as the U.S. FDA and the European Medicines Agency (EMA), or relevant specialized groups such as CLSI-AST and the European Committee on Antimicrobial Susceptibility Testing (EUCAST), in defining the susceptibility breakpoints [15].
The purpose of this investigation was to study the PK/ PD relationship of marbofloxacin against H. parasuis, derive a CO PD of marbofloxacin and recommend a reasonable dosage regimen. MICs of marbofloxacin against those isolates were determined. An in vitro PK/PD infection model was used to investigate marbofloxacin effects against H. parasuis strain of serotype 5, which is a highly virulent serotype and is one of the most prevalent serotypes in China. Pharmacokinetics of marbofloxacin in swine obtained from a previous study and PK/PD indices were integrated into a Monte Carlo simulation to derive a CO PD . A rational regimen of marbofloxacin against H. parasuis was determined.

Animal ethics
All husbandry practices and experimental operations were performed with full consideration of animal welfare. Research ethical approval was granted by the South China Agriculture University Animal ethics committee (2014-03).

Strains and antibiotic
A strain of H. parasuis, serovar 5 (V5), kindly provided by Professor Ming Liao, College of Veterinary Medicine, South China Agricultural University, Guangzhou, Guangdong Province, China was used in the present study. V5 was marbofloxacin susceptible with a MIC of 0.015 mg/L. The other 189 isolates of H. parasuis were collected from swine in south China regions between August 2010 and July 2011. Serotypes of these isolates are not known. Tryptone soya agar (TSA) and Tryptone soya broth (Oxoid Ltd., Basingstoke, Hampshire, UK), supplemented with 2 % beta-Nicotinamide adenine dinucleotide trihydrate (NAD) (Qingdao Hope Bio-Technology Co., Ltd., Shandong, China) and 5 % new-born calf serum (Guangzhou Ruite Bio-tec Co., Ltd., Guangdong, China), were used to culture H. parasuis. Marbofloxacin was purchased from Hebei Yuanzheng Pharmaceutical Company (Hubei, China).

Susceptibility testing
Considering that there is no recommended testing method by CLSI's VAST Subcommittee for H. parasuis, MICs were conducted in accordance with the CLSI recommendations for Actinobacillus pleuropneumoniae.
[16] (Wakiec et al., 2008) The Actinobacillus pleuropneumoniae ATCC 27090 strain was used for quality control purpose. MIC 50 and MIC 90 were defined in the present study as the lowest concentration that inhibited the growth of 50 % and 90 % of isolates tested, respectively. The mutant prevention concentration (MPC) value was determined according to previous reports [17,18]. Single bacteria colony from 24 h growth on TSA was grown for 12 h in TSB broth, then concentrated through centrifugation and re-suspended it in TSB to a final concentration of~3 × 10 10 CFU/mL. An aliquot of 500 μL samples was plated onto TSA plates containing various concentrations of marbofloxacin, and then incubated for 24 h, 36 h and 48 h for re-growth. MPC pr was defined as the lowest drug concentration that inhibits growth. A second measurement was performed using linear drug concentration increment within 20 % per sequential increase. All the determinations were carried out in triplicates.

In vitro PK/PD model
The in vitro one-compartment PK/PD infection model equipment was constructed according to previously described method with some improvements [19]. An inverted 50 mL centrifuge tube with a cellulose ester membrane (0.2-μm pore size) covering the top was placed in the central compartment to prevent bacteria from flowing out to the medium. A magnetic stir bar was placed on the bottom of the central compartment which mixed the broth and enabled the drug to fully contact the bacteria. The flow rate was 0.171 mL/min to simulate the half-life of marbofloxacin in swine as described previously [20].

In vitro time kill curves of marbofloxacin
A 12 h culture of V5 at logarithmic phase was added to the central compartment to reach a final concentration of 10 7 colony forming unit (cfu)/mL. An incubation period of 30 min was applied to adapt the bacteria to the new environment. Different doses of marbofloxacin or control (sterile normal saline) were administered into the central compartment, and at the same time, the peristaltic pump was turned on. Samples were obtained at time points of 0, 3, 6, 9, 12, and 24 h. 100 μL of the samples were diluted properly with sterile normal saline, aliquots of the last four diluted samples were dropped onto the TSA plates and incubated at 37°C for 24 h. The limit of determination was 400 cfu/mL.

Pharmacokinetics and PK/PD analysis
The samples for marbofloxacin concentration determination were centrifuged at 8,000 rpm at 4°C for 10 min. The supernatant was stored at −80°C and analyzed using High Performance Liquid Chromatography (HPLC), which had been optimized by our laboratory [21] within 1 month. All experiments were performed in duplicate on different days. The PK data were analyzed using Phoenix WinNonlin 6.0 software (Pharsight Co. Ltd.).
As marbofloxacin is concentration dependent, the PK/ PD index of marbofloxacin was AUC 24h /MIC and C max / MIC [22]. The PK/PD indexes were calculated using the pharmacokinetic data and MIC value in each dose of the time-kill curve. The in vitro drug effect was quantified by changes in log 10 cfu counts between 24 h and 0 h. Data were analysed using sigmoid E max model WINNONLIN software (version 6.1; Pharsight, CA, USA) with the following equation: Where E 0 is the change in log 10 cfu/mL after 24 h incubation in the control sample, compared with the initial inoculum. E max is the difference in effect between the greatest amount of growth (as seen for the growth control, E 0 ) and the greatest amount of kill. C e is the AUC 24h /MIC, C max /MIC or C max /MPC in the effect compartment. EC 50 is the AUC 24h /MIC, C max /MIC or C max /MPC value producing a 50 % reduction in bacterial counts from the initial inoculum, and N is the Hill coefficient that describes the steepness of the curve.

Monte Carlo analysis (MCS)
A 10,000-subject Monte Carlo simulation was conducted using Crystal Ball Professional V7.2.2 software based on a previous pharmacokinetic study of marbofloxacin in pigs and PK/PD target indices obtained in this study. AUC 24h and C max were assumed to be log-normally distributed in the form of mean values and confidence intervals. The CO PD is the MIC at which the probability of target attainment (PTA) equals to 90 %, which is the most commonly used standard for susceptibility breakpoints in other bacteria [12].

Dosage calculation
In order to deduce a more rational regimen, the general formula was employed to estimate dosages for different magnitudes of efficiency [23].
Where Dose is the optimal dose (mg/kg day), CL is the body clearance (L/kg day), AUC/MIC is the breakpoint marker for the desired effect, MIC 90 is the MIC inhibiting 90 % of strains (mg/L), F is the bioavailability, and fu is the free drug fraction.

Results
MICs of marbofloxacin against H. parasuis were widely distributed, ranging from 0.003 mg/L to 16 mg/L (Fig. 1). A trimodal distribution was observed with peak value observed at 0.003, 0.06, and 2 mg/L, respectively. The MIC 50 and MIC 90 were 2 and 8 mg/L, respectively. The MPC of strain V5 was 0.04 mg/L at different time endpoints (24, 36 and 48 h), which was about 2~3 times of the MIC value (0.015 mg/L).
The pharmacokinetics of marbofloxacin in pigs was well simulated by this model with the relative deviation below 7 %. Time killing curves were shown in Fig. 2. The marbofloxacin inhibited H. parasuis moderately when the AUC 24h /MIC was less than 13.8. The bacteria decreased rapidly within 12 h, but re-grew to 10 6 cfu/ mL at 24 h with the AUC 24h /MIC of 46. When the AUC 24h /MIC was 55 and 73, H. parasuis could not be detected at 12 h, however, re-growth was observed at 24 h. When AUC 24h /MIC was 92 and 110, marbofloxacin killed H. parasuis without regrowth in 24 h (4-log-unit and 5-log-unit decrease, respectively). The relationship between in vitro antimicrobial efficacy and PK/PD surrogate markers (AUC 24h /MIC or C max /MIC) was described using the sigmoid E max model. Both of the PK/PD surrogate markers simulated the in vitro antimicrobial effect of marbofloxacin successfully with the R 2 of 0.9928 and 0.9911 respectively (Figs 3, 4). The estimated Log E max , Log E 0 , EC 50, and slope were shown in Tables 1 and 2 respectively. The target values of 3-log 10 -unit and 4log 10 -unit decreases for C max /MIC were 6.5 and 8 while for AUC 24h /MIC were 88 and 110, respectively. The same effects for surrogates C max /MPC and AUC 24h / MPC were 2.5, 3 and 33, 42 respectively. They also simulated the in vitro antimicrobial effect of marbofloxacin successfully with the R 2 of 0.9928 and 0.9911 respectively.
As there are no PK data of marbofloxacin with a dose of 2 mg/kg BW, data of 2.5 mg/kg BW were used for MCS. In the simulation, with the PK/PD target AUC 24h / MIC of 88, PTA > 90 % could only be achieved for isolates with MIC ≤ 0.125 mg/L (Fig. 5). For dosage regimen of 8 mg/kg BW with a single dose administered by IM, PTA > 90 % could be achieved for isolates with MIC ≤ 0.5 mg/L (Fig. 5). The CO PD for marbofloxacin against H. parasuis was 0.5 mg/L. The recommended dose of marbofloxacin for bactericidal effect against H. parasuis was 16 mg/kg BW.

Discussion
Currently, a great deal of information is available on the PK/PD relationship of fluoroquinolones. The parameters C max /MIC and AUC 24h /MIC correlate well with therapeutic outcome. According to literature, AUC 24h / MIC of >125 h and C max /MIC of >10 was usually used as a threshold for successful therapeutic outcome of fluoroquinolones against gram negative bacteria [23]. Nevertheless, these thresholds may be different for some fluoroquinolones. The greatest influence for the differences was the immune status of the animal. Furthermore, the PK/PD indices of the same drug against different pathogens also vary. For example, the threshold of AUC 24h /MIC is 46 h for bactericidal action in an ex vivo PK/PD study of marbofloxacin against Mannheimia haemolytica [24]; AUC 24h /MIC ratios for no reduction, 3 log10 and 4 log10 reductions in bacterial count from the initial inoculum count were 41.9, 59.5, and 68.0 h for M. haemolytica and 48.6, 64.9, and 74.8 h for P. multocida in an ex vivo PK/PD study of marbofloxacin [25]. So it is of great importance to study the PK/PD indices of fluoroquinolones individually. Data on the PK/PD indices of marbofloxacin against H. parasuis are limited. In this study, PK/PD  The AUC 24h /MIC ratios for no reduction, 3 log10, and 4 log10 reductions in bacterial count were 50, 88, and 110 h, while the C max /MIC ratios for those effects were 3.5, 6.5, and 8. The threshold value is higher than those derived from ex vivo PK/PD model. For these comparisons, it should be noted that there are significant differences between dynamic in vitro and ex vivo conditions. There is a continuous exposure to a fixed concentration of the agent for a defined duration (e.g., 24 h) in an ex vivo model, whereas a gradient of concentration in the dynamic in vitro model. The level of AUC 24h /MPC (33 for 3 log 10 reductions) was higher than that (12.89 for bactericidal effect) resulted from a tissue cage model of marbofloxacin against Pasteurella multocida [26]. The greatest influence for the differences might be the immune status of the animal. The value of C max /MPC (2.5 for bactericidal effect) was almost the same with that (C max /MPC > 2.2) of levofloxacin against staphylococcus aureus in a hollow fiber PK/PD model [27]. Though, both the PK/PD surrogates derived from MIC and MPC described the effect properly, PK/PD surrogates derived from MPC has been proven to be superior for fluoroquinolones over the classic PK/PD indices based on MIC for minimizing the emergence of resistances and preventing therapeutic failure [27].
Susceptibility breakpoint setting requires knowledge of the wild-type distribution of MICs, assessment of the PK/PD indices, and study of the clinical outcome of infections treated with the antibacterial. Monte Carlo simulation shows great advantage in supporting determination of the susceptibility breakpoint using drug exposure-effect relationship [14], which takes pharmacokinetic variation and PK/PD indices into consideration. For the simulation, as it had been proved that the efficacy of a single dosing regimen was better than doses administered every 24 h or 48 h of the same total amount of marbofloxacin [26], the EMA recommended dose regimen can be converted to 8 mg/kg BW with a single dose. Though both AUC 24h /MIC and C max /MIC were PK/PD indices of fluoroquinolones, AUC 24h is a much more robust estimation than the one of a single snapshot C max which depends on many factors such as sampling schedule. Therefore the PK/PD target was defined to be AUC 24h /MIC with value of 88. The CO PD of marbofloxacin against H. parasuis was 0.5 mg/L under one short dose of 8 mg/kg. This CO PD value was equal to the clinical breakpoints values and the PK/PD breakpoints values of ciprofloxacin, moxifloxacin (0.5 mg/L), ofloxacin (0.5 mg/L), and levofloxacin(1 mg/L) against Haemophilus influenzae, another bacteria of Haemophilus spp in human clinic use [28]. Unfortunately, the proposed PK/PD cutoff would designate a large proportion of clinical H. parasuis isolates as resistant to the marbofloxacin. Swine infected by these isolates will probably have a low likelihood of responding to therapy.
Clinical dosage regimens of antimicrobial agents are traditionally determined by relating the PK of drugs in healthy animals and the in vitro antibacterial activity or the treatment outcome at given dosages in disease models or in clinical trials involving limited strains. Although these parameters can predict the potency of the drug against pathogens to a certain extent, they actually Note: E 0 is the change in log 10 cfu/mL after 24 h incubation in the control sample compared with the initial inoculum. E max is the difference in effect between the greatest amount of growth (as seen for the growth control, E 0 ) and the greatest amount of kill. EC 50 is the C max /MIC value producing a 50 % reduction in bacterial counts from the initial inoculum, and N is the Hill coefficient that describes the steepness of the dose-response curve Note: E 0 is the change in log 10 cfu/mL after 24 h incubation in the control sample compared with the initial inoculum. E max is the difference in effect between the greatest amount of growth (as seen for the growth control, E 0 ) and the greatest amount of kill. EC 50 is the AUC 24h /MIC value producing a 50 % reduction in bacterial counts from the initial inoculum, and N is the Hill coefficient that describes the steepness of the dose-response curve don't provide information on the time course of antimicrobial activity. Fortunately, the relationship of PK/PD parameters and the clinical outcomes have been fully investigated and applied in optimizing drug regimens; the regimens based on PK/PD results were usually calculated by the general formula [23]. In the calculation, the PK/PD threshold, the PK parameters and MIC distribution were taken into accounted. As the fu of marbofloxacin in swine plasma was sparse, the value in dog plasma was used to substitute that. According to our best knowledge, the CL of marbofloxacin was 0.065 ± 0.011 L/kgh; the bioavailability was 90 % ± 28 %; the protein binding rate was 21.81 % ± 6.26 %; MIC 50 is 2 mg/L [20,29]. When the MIC 90 was used to calculate the dose, we found the result (64 mg/kg BW) was too high to apply in clinical usage. Considering the MICs of some H. parasuis isolated in China are very high, they may exhibit resistance to marbofloxacin that cannot be treated by marbofloxacin. MIC 50 was used in the calculation. the result showed that a dose of 16 mg/kg would be able to achieve bactericidal effect. This dose is much higher than the recommended dose (2 mg/kg for 3 to 5 days), and higher than that (10 mg/kg) [30] recommended for bovine respiratory disease based on the theoretical principle of single-injection, short-acting antibiotic (SISAAB), which has been primarily developed and applied to fluoroquinolones used in human medicine [31]. It seems that even a dose as high as 16 mg/kg BW cannot cure all the H. parasuis infection in China. It is well-known that fluoroquinolones can lead to crossresistance among different members of the class [32], since they are widely used to treat respiratory diseases. It is better to check the susceptibility of pathogens before drug administration, as resistance determinants may transfer to human pathogenic bacteria, resulting in the failure of antibiotics in treatment of bacterial infection. However, there are some limitations in our study. First, the PK/PD indices targets were based on the drug concentration at infection sites, whist drug concentration of plasma were used for MCS. Though, the penetration of marbofloxacin is good and the tissue concentration is similar to that in blood [26], concentration at infection sites should be simulated in future trials. A second limitation is that the recommended regimen is useful in just the regions of China as MIC probability distribution of a determined pathogen may vary between countries and regions and even time. Finally, our proposed CO PD will need to be validated in the clinical outcome.

Conclusions
In summary, this study established an in vitro dynamic PK/PD modelling of marbofloxacin against H. parasuis. The target PK/PD values of marbofloxacin for 3-log 10unit and 4-log 10 -unit decreases effects were C max /MIC of 6.5 and 8, C max /MPC of 2.5 and 3, AUC 24h /MIC of 88 and 110 or AUC 24h /MPC of 33 and 42 respectively. The very first marbofloxacin CO PD (0.5 mg/L) derived based on MCS was of great utility in marbofloxacin susceptibility test and dosing design. Marbofloxacin can have the best efficacy at dosage of 16 mg/kg BW for strains with MIC values ≤ 2 mg/L, therefore, it is obligatory to know the sensitivity of the pathogen and to treat animals as early as possible.