From March to May 2014, a total of 103 H. parasuis strains were donated by State Key Laboratory of Agricultural Microbiology at Huazhong Agricultural University and National Reference Laboratory of Veterinary Drug Residues at South China Agricultural University. These strains were isolated from the lungs of swine in 10 provinces of China. All of the bacterial isolates were confirmed by polymerase chain reaction (PCR). Prior to testing, each isolate was subcultured at least twice on tryptic soy agar (TSA) containing 5% fetal calf serum (FCS) and 10 μg/mL nicotinamide adenine dinucleotide (NAD) to ensure viability and purity.

Susceptibility testing was performed by the agar dilution method according to the CLSI M07-A9 standard with some modification based on the characteristics of H. parasuis. A 2 μL H. parasuis suspension (10⁸ CFU/mL) was inoculated onto TSA-FCS-NAD agar plates containing twofold dilutions (0.015~32 μg/mL) of tilmicosin (Dr. Ehrenstorfer Standards, Augsburg, Germany). Plates were incubated at 37°C in an atmosphere containing 5% CO₂ for 36 h. Enterococcus faecalis (ATCC 29212) was used as the quality control (QC) strain to ensure the credibility of MICs tested.

A microorganism is defined as WT for a species by the absence of resistance mechanisms to target drug (Turnidge et al., 2006; Turnidge and Paterson, 2007). The ECV, or CO_WT is used to separate bacterial populations on the basis of MIC distributions (Turnidge and Paterson, 2007; Espinel-Ingroff et al., 2010; Canton et al., 2012). Ideally, at least 95% of WT isolates should be encompassed into the ECV (Pfaller et al., 2009, 2010, 2011). The ECV was calculated following the method described by Turnidge (Turnidge et al., 2006). Briefly, Normality testing of the WT distribution was conducted with Sigmastat software v.3.5, non-linear regression was used to fit log₂-transformed MICs with Graphpad Prism v.5.01, NORMINV and NORDIST functions were employed to set the WT distribution cutoffs.

The molecular mechanism involved in macrolide resistance in H. parasuis is unclear, however, mutations in the 23S rRNA that is commonly associated with macrolide-resistance may occur in H. parasuis. Therefore, PCR amplification (Hao et al., 2013) with the primers 23S-F (5′-ACGGTCCTAAGGTAGCGAAAT-3′) and 23S-R (5′-CATCAAATGTTAAAGGGTGGTA-3′) was used to screen mutations in the 23S rRNA gene in H. parasuis.

According to the determined MIC of H. parasuis SH0165, agar plates were prepared with concentrations of tilmicosin (Dr. Ehrenstorfer standards, Augsburg, Germany)) ranging from 1/4 to 32 MIC. A total of 0.1 mL of inoculum was plated and bacterial counts were performed at 0, 1, 2, 4, 8, 12, and 24 h.

Fourteen 10-weeks-old healthy crossbred (Duroc × Large white × Landrace) pigs weighing 28–33 kg were purchased from Huazhong Agricultural University pig breeding farm. Prior to experiments, pigs were raised 7 days to acclimate. Two pigs were used for establishment of high performance liquid chromatography (HPLC) method, another 12 pigs were used for pharmacokinetics (PK) study.

All the animal experiments were approved by the Animal Ethics Committee of Huazhong Agricultural University (hzauch 2014-003) and the Animal Care Center, Hubei Science and Technology Agency in China (SYXK 2013–0044). All efforts were used to reduce the pain and adverse effect of the animals.

The 12 pigs were randomly divided into two groups. Tilmicosin was orally administrated to six pigs in each group at a single-dose of 40 mg/kg/bw. This dose was designed based on the clinical outcome of 400 mg/kg feed (about 40 mg/kg/bw) for treatment of H. parasuis in pigs (MacInnes et al., 2003). In Group A, blood (2 mL) samples from animals were obtained at 0, 0.5, 0.75, 1, 2, 3, 4, 5, 6, 8, 12, 24, 36, 48, 72, and 96 h after administration. In Group B, PELF samples were collected at 0, 0.5, 2, 4, 6, 8, 12, 24, 36, 48, 72, and 96 h post-drug administration.

Atropine (0.05 mg/kg), ketamine (5 mg/kg), and propofol (3 mg/kg) were given intramuscularly and intravenously 30 min before drug administration.

Standardized BAL was performed as previously described (Yamazaki et al., 2003; Choi et al., 2012; Lee et al., 2015), with an electronic fiberoptic bronchoscope (Kangmei GU-180VET) inserted in the right middle lung lobe. Then, 50 mL of normal saline was instilled into the lobe, and was aspirated into a 50 mL centrifugal tube after 20 s.

Plasma was separated from blood by centrifugation at 3000 × g for 10 min and was kept at -70°C until assay. The PELF was centrifuged at 400 × g for 10 min and stored at -70°C until analysis.

Quantitation analysis of tilmicosin in PELF and plasma were conducted using HPLC. A C₁₈ reverse-phase column (250 mm × 4.6 mm i.d., 5 μm; Agilent) was used to perform HPLC at 30°C. The detection wavelength was 285 nm. The mobile phase consisted of 5 mM ammonium formate added with 0.1% formic acid (phase A) and acetonitrile (phase B; 73:27, v/v).

Plasma (0.5 mL) was extracted with dichloromethane (2.5 mL) twice. After centrifugation, supernatant was evaporated and resuspended in the mobile phase (0.5 mL). PELF (0.5 mL) was extracted with acetonitrile (2 mL), and then centrifuged, evaporated and resuspended in a same manner.

The limit of determination (LOD) was 0.02 μg/mL and the limit of quantification (LOQ) was 0.05 μg/mL both in plasma and PELF. Standard curves were linear from 0.05 to 10 μg/mL both in plasma (R² = 0.9998) and PELF (R² = 0.9999). The inter-day variation for determination in plasma ranged from 0.78 to 1.14% and PELF 0.14 to 0.78%, respectively. The recovery of tilmicosin in plasma ranged from 93.67 ± 1.06% to 97.16 ± 1.00% and in PELF from 98.49 ± 0.77 to 99.44 ± 014%.

The urea dilution method was used to determine the volume of PELF as described previously (Conte et al., 2000a; Kiem and Schentag, 2007). The concentration of urea in plasma and PELF were determined by the urease-glutamate dehydrogenase enzymatic method with an automatic biochemical analyzer (SYNCHRON CX4 PRO; Beckman) at the National Reference Laboratory of Veterinary Drug Residues (Wuhan, China).

Statistical analysis was conducted by using WinNonlin v. 5.2.1. Plasma concentration data was analyzed with a two-compartment model and PELF non-compartment model according to the characteristics of concentration-time data.

Crystal Ball v7.2.2 was used to perform Monte Carlo simulation. The mean value and standard deviation of AUC₂₄ of PELF were embedded in the function. The distribution of pharmacokinetic parameter AUC₂₄ was assumed to be log-normal. A total of 10000 subjects were simulated.

The PK/PD target was evaluated (Andes et al., 2004). Conservative value (AUC/MIC = 30) was selected to calculate the probability of target attainment (PTA). CO_PD was defined as the MIC at which the PTA was ≥90%.

All presumptive H. parasuis isolates were confirmed by PCR-based method. The MIC for Enterococcus faecalis (ATCC 29212) was 8 μg/mL, which was within the acceptable QC range according to CLSI document M31-A3.

The WT MIC distribution for tilmicoin against H. parasuis is shown in Figure 1. The tilmicosin MIC ranged from 0.06 to 16 μg/mL. The distribution of each MIC (0.06, 0.125,0.25, 0.5, 1, 2, 4, 8, and 16 μg/mL) among tested isolates was as follows: 0.06 μg/mL (1.94%), 0.12 μg/mL (4.85%), 0.25 μg/mL (19.42%), 0.5 μg/mL (15.53%), 1 μg/mL (28.16%), 2 μg/mL (15.53%), 4 μg/mL (10.68%), 8 μg/mL (2.91%), and 16 μg/mL (0.97%). The MIC₅₀ and MIC₉₀ were 1 and 4 μg/mL, respectively.

Despite a lower prevalence at the MIC of 0.5 μg/mL, cumulative counts of MIC data were found to match a good-shape normal distribution by use of normality test (P = 0.200) as shown in Figure 2. The optimum MIC range (0.015–16 μg/mL) was obtained using non-linear regression (Table 1). In addition, this range was further corrected to 0.06–16 μg/mL by employing the NORMINV function. The probabilites of an isolate MIC value higher than the high cutoff (0.18%) and lower than the low cutoff (0.07%) were estimated using NORMDIST function. As the result, the ECV was defined as 16 μg/mL, which encompassed 100% of the WT isolates (Table 1).

After PCR amplification and DNA sequencing, no 23S rRNA gene mutations associated with macrolide resistance were found among all of the tested H. parasuis isolates. Therefore, all isolates were assumed to be WT strains.

As displayed in Figure 3, the lower concentrations (≤MIC) of tilmicosin exhibited similar antimicrobial activity to H. parasuis. However, when tilmicosin concentrations were higher than MIC, the bacteriostatic efficiency gradually strengthened due to increased drug concentration. Therefore, the in vitro time kill curve showed that activity of tilmicosin against H. parasuis was concentration-dependent.

No serious adverse influences were observed after oral administration of tilmicosin. The concentration of tilmicosin in plasma was reduced below the LOQ after 72 h. A two-compartment model was used for model fitting from 0 to 72 h. The concentration and time profiles are illustrated in Figure 4. Pharmacokinetic parameters of tilmicosin in plasma were calculated by model analysis conducted by WinNonlin v. 5.2.1. As shown in Table 2, the time to reach to maximum concentration (Tmax), the peak drug concentration (Cmax), and the area under the curve at 24 h (AUC₂₄) were 3.52 ± 0.34 h, 1.57 ± 0.46 μg/mL, and 20.13 ± 5.26 μg. h/mL, respectively. The mean residence time (MRT) was 16.45 ± 1.67 h. The intercept for the distribution phase (A) and for the elimination phase (B) were 2.67 ± 0.99 μg/mL and 0.09 ± 0.01 μg/mL. The distribution rate constant (α) and elimination rate constant (β) were 0.11 ± 0.01 L/h and 0.002 ± 0.001 L/h.

After drug administration, the tilmicosin concentration in PELF reached to a higher level at 3 h and kept at that level until 96 h. A non-compartment model was selected to analyze the drug concentration and time characteristics for PELF samples. The pharmacokinetic parameters for PELF samples are summarized in Table 2. The values (mean ± SD) of T_max, C_max, AUC₂₄, and MRT were 40.80 ± 6.57 h, 5.36 ± 0.74 μg/Ml, 74.41 ± 17.98 μg.h/mL, and 37.64 ± 1.86 h, respectively.

The concentration-time curves both in plasma and in PELF after oral administration of tilmicosin at a single dose of 40 mg/kg/bw are shown in Figure 4. The drug concentration in plasma and in PELF was lower than 2 μg/mL during 0~4 h. The concentration of tilmicosin in plasma rapidly decreased from 1.36 μg/mL at 6 h to 0.11 μg/mL at 72 h. In contrast, the drug concentration in PELF remained more stable, as values were higher than 4 μg/mL from 6 to 48 h and then showed a rapid decrease from 4.72 μg/mL at 48 h to 0.68 μg/mL at 96 h. Significant differences were observed between drug concentrations in plasma and in PELF. It was noteworthy that the drug concentration in PELF at 48 h reached 4.72 ± 0.44 μg/mL, which was more than 40 times of that (0.11 ± 0.03 μg/mL) detected in plasma. Both the values for C_max and AUC₂₄ in PELF were obviously higher than those in plasma (Table 2).

As presented in Table 3, 10,000 subjects were modeled by employing Monte Carlo simulation. The PTA under different WT MICs was calculated. The PTA achieved to 100% when MIC value of 1 μg/mL was employed. However, the PTA was merely 76% (far below 90%) under the MIC value of 2 μg/mL. Consequently, the CO_PD was defined as 1 μg/mL.