| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
ANIMAL PRODUCTION |
* UMR 181 de physiopathologie et toxicologie expérimentales INRA/ENVT, 23 Chemin des Capelles, F-31076, Toulouse cedex 3, France;
and
Centre Technique des Productions Animales et agro-alimentaires, Rond point du Zoopole, BP 7, F-22440 Ploufragan, France; and
and
Laboratoires Sogéval. 200, avenue de Mayenne, BP 2227, F-53022 Laval cedex 9, France
 |
Abstract |
|---|
 Top Abstract INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONIMPLICATIONSLITERATURE CITED |
|---|
Key Words: acetaminophen • doxycycline • feed supplement • pharmacokinetics • pig • respiratory disease
|
INTRODUCTION |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONIMPLICATIONSLITERATURE CITED |
|---|
). Because bacteria are involved in this condition, antibiotics are a short-term measure in a multifaceted strategy of disease control.
Tetracyclines are the most commonly used antibiotics in animal health. Feed is the most efficient vehicle for antibiotic therapy in swine herds, but the clinical efficacy of feed medication depends on feeding behavior (del Castillo et al., 1998). In systemically infected pigs, endogenous and bacterial anorexigenic substances decrease appetite by increasing PGE2 concentration in cerebrospinal fluid.
Paracetamol (PARA) inhibits the intracerebral synthesis of PGE2, shows potent antipyretic and analgesic effects, and improves feed intake in laboratory animals exposed to endotoxin. The appetite-stimulating effect of PARA in infected swine has not been examined. But it likely exists because its mode of action resembles that of indomethacin, which prevented endotoxin-induced anorexia and fever in pigs (Johnson and von Borell, 1994). We have noted that in-feed PARA seems to improve the clinical efficacy of in-feed doxycycline (DOX) against PRDC (C. Zémirline, unpublished data). We hypothesize that PARA selectively increases the systemic exposure to DOX in infected pigs by improving their appetites to a level that is closer to that of healthy pigs. Our objectives were to document the pharmacokinetics (PK) of DOX in pigs, when it is coadministered with PARA intravenously and in liquid feed, to compare the systemic exposure to in-feed DOX in groups of pigs naturally infected with bacterial PRDC when given alone (DOX) or combined with PARA (PARADOX) in liquid feed. Systemic exposure to DOX were measured with the area under the curve (AUC) of its plasma concentrations over time. A third objective was to document interindividual variability in AUC to in-feed DOX in swine herds.
|
MATERIALS AND METHODS |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONIMPLICATIONSLITERATURE CITED |
|---|
Experiment 1: Individual Pharmacokinetics of DOX and Development of the Optimal Sampling Strategy for Population Studies
Animals, Housing, and Surgery.
This experiment was performed at SOURCHES Agro-Evolution research farm (Saint-Symphorien, France). We used 5 healthy Large White-Landrace x Piétrain pigs (2 males, 3 females). Pigs were 8 wk old and 18 kg of BW when they arrived at the facility, after which they were allowed 7 d to acclimatize. Pigs were placed in 1.2 m2 individual cages with 2/3 concrete, 1/3 wire-mesh decks; room temperature was set at 25°C; and daylight was the main source of light. Drinking water was offered ad libitum, and 2 meals of 500 g of a drug-free diet (Porc croissance, Glon-Sanders, Pontivy, France; 13 MJ/kg of ME, 16% CP), mixed with 1,200 mL of tap water immediately before serving, were given daily at 0800 and 1630. An extra drug-free meal (200 g of feed + 480 mL of water) was offered at 1330 on the days the PK experiment took place.
On d 5 of their stay in the facilities, pigs were anaesthetized with 4 mg of azaperone (Stresnil, Janssen-Cilag SA Division Santé Animale, Issyles-Moulineaux, France)/kg (i.m.), 25 mg of ketamine (Imalgene, Mérial, Lyon, France)/kg (i.m.), and 2 mg of propofol (Rapinovet, Schering-Plough Vétérinaire SA, Levallois-Perret, France)/kg (i.v.). Ketoprofen, 3 mg/kg (i.m.) was given for postoperative analgesia (Kétofène 10%, Mérial, Lyon, France). Indwelling catheters (Vygon, Laboratoires Vygon, Ecouen, France) were implanted surgically in the left and right external jugular veins (for sampling and dosing, respectively). An adhesive elastic bandage fixed the catheters, and the free end (30-cm long) allowed repeated blood sampling on the unrestrained pigs. Catheters were filled with 0.9% NaCl sterile solution containing 100 IU of heparin/mL (Sanofi-Synthélabo, Le Plessis Robinson, France).
Drugs.
Sterile, 10 mg/mL solutions of DOX (Doxycycline Hydrochloride, Sigma-Aldrich 
himie, St-Quentin Fallavier, France); Lot #100K1340, potency = 864 mg/g) and PARA (Acetaminophen, Sigma Chemical Co.; Lot #060H0801, potency > 990 mg/g) in 0.9% NaCl solution were prepared for i.v. dosing. Dissolution of PARA required the addition of 4% vol/vol ethanol. Solutions were sterilized with 0.20-µm syringe filters. Feed-grade premixes of DOX (Doxyval 40 Porc P/M. Lot #2105101; potency = 41.5 mg/g) and PARA (Concentrat V079 Paracétamol 100 Porc. Lot #2108437; potency = 96.3 mg/g) were provided by Laboratoires Sogéval, Laval, France.
Pharmacokinetic Experiment.
On d 8, pigs were weighed and their respective morning meals were fortified with DOX and PARA premixes to an amount corresponding, respectively, to 10 and 30 mg/kg of BW. Blood samples were collected from the right catheter at 0, 0.5, 1, 1.5, 2, 3, 4, 5, 6, 9, 12, 14, and 24 h after the medicated meal was offered.
On d 9, the pigs were weighed before their morning meal, the 24-h blood sample was collected, and 10 mg of DOX and PARA/kg of BW were injected i.v. through the left catheter. We used this low dose of PARA to minimize the side effects of ethanol (the diluter of i.v. PARA), assuming that the PK of PARA is linear (Bannwarth and Péhourcq, 2003) and that PARA is unlikely to produce metabolic interactions with DOX, which does not seem to be metabolized in swine (Riond and Riviere, 1990). Blood samples were collected at 0.1, 0.2, 0.4, 0.6, 0.8, 1, 2, 3, 4.5, 6, 9, 12, 24, 27, and 30 h after injection. Blood samples were transferred immediately to vacuum tubes containing lithium heparin (Vacutainer, Becton Dickinson, Plymouth, UK) and centrifuged for 10 min at 1,500 x g. Plasma was divided into 4 aliquots (2 for analyzing DOX concentrations, and 2 for PARA concentrations) and stored at –20°C pending analysis.
Pharmacokinetic Analysis.
Total plasma concentrations of DOX and PARA were measured using separate HPLC methods, with lower limits of quantification and precisions of 0.05 mg/L and <15% for DOX, and 0.10 mg/L and <10% for PARA (Phatophy SARL, Lyon, France). Subsets of plasma concentrations of DOX associated with in-feed and i.v. dosing in each animal (i.e., from 0 to 24 h, and from 24 to 54 h, respectively) were subjected to separate statistical-moment PK analyses (Yamaoka et al., 1978). The AUC of DOX plasma concentrations and that of the product of time and DOX concentration (i.e., the area under the first moment curve; AUMC) were calculated with WinNonlin (Pharsight Corp., Mountain View, CA) using the linear/log trapezoid rule (Purves, 1992). The trapezoids of in-feed DOX from the time of i.v. dosing to infinity were subtracted from the AUC and AUMC of i.v. administered DOX to avoid bias when estimating its PK disposition parameters. In addition, we determined the slope of the terminal decay processes (
z) of in-feed and i.v. DOX, its observed peak plasma concentration (Cmax) after in-feed dosage, and the time of Cmax occurrence (tmax).
These estimates of the PK parameters were used as initial values for fitting the whole set of individual plasma concentrations of DOX (i.e., from 0 to 54 h) to an open 2-compartment model, where BW is a covariate for the volume of the central compartment, which is connected to an absorption compartment with 2 output rates. This model was preferred to others used in simultaneous fitting of oral and i.v. kinetic profiles (e.g., models with 1 or 3 disposition compartments) based on its accuracy in parameter estimation, the Schwarz goodness of fit criterion, and visual inspection of residuals. We numerically solved the linear differential equations of this compartmental model with ADAPT II (Biomedical Simulation Resource, University of Southern California, Los Angeles), using maximum likelihood estimation of PK parameters (D’Argenio and Schumitzky, 1997).
The volume of the central compartment (Vc) was deduced from the formula Cp = X1/(Vc·W), where X1 is the amount of DOX in the central compartment, Cp is the plasma concentration of DOX, and W is BW prior to DOX administration. The systemic clearance of DOX (CLt) was calculated with CLt = Vc·k10, where k10 is the elimination rate constant from the central compartment. The distribution clearance (CLd) was calculated with CLd = Vc·k13, where k13 is the transfer rate constant from the central to the peripheral compartment. The steady-state volume of distribution (Vss) was calculated with
 |
where k31 is the transfer rate microconstant from the peripheral to the central compartment. The oral bio-availability (F) of DOX was defined as F = ka/(ka + kp), where ka and kp are the absorption and the presystemic elimination rate constants, respectively, from the absorption site.
Mathematical Simulations.
Repeated blood sampling is required to ascertain the systemic exposure to DOX, but on farms this can only be obtained through harmful venipuncture. To ensure the well being of the pigs and the quality of information, we generated a sparse-sampling strategy with optimal blood collection times for obtaining the most accurate AUC estimates of in-feed DOX. Therefore, individual estimates of systemic exposure to DOX were obtained with the lowest number of punctures on each pig.
Devising the optimal sparse-sampling strategy for a farm-based PK experiment requires prior knowledge of the population PK parameters. Because these were unknown, we used the empirical means and variances of our individual PK parameters as preliminary population estimates (i.e., a 2-stage method). For all sparse-sampling designs to be examined, sets of individual PK parameters were drawn from log-normal distributions to simulate the time-course of DOX concentrations in plasma with the Bateman function. These simulated sparse data were analyzed with a nonlinear mixed-effects model to obtain estimates of the population PK parameters. The mean square error of the population average AUC estimate was the criterion used to choose the optimal sparse sampling design.
All sparse-sampling designs considered in this study consisted of 2 fixed times, starting 33 min before serving the second medicated meal and 24 h afterwards, plus 4 times to distribute optimally within 12 h after the beginning of the second medicated meal. The time required for puncturing all the test pigs in a time was set at 30 min. The minimum interval between the starting times of 2 consecutive times was set at 10 min. The time starting 24 h after serving the second medicated meal was used only in 25% of pigs chosen randomly in each pen for each treatment group.
Experiment 2: Population Pharmacokinetics of DOX and PARADOX
The orexigenic effect of PARA cannot be detected upon its first administration, and the appetite of diseased pigs treated with DOX alone may increase with time and mask the expected difference between the 2 in-feed dosage regimens. Hence, we assumed the difference in systemic exposure was more likely to be detected early in the course of the feed dosage regimen. Therefore, our farm-based PK experiment compared the systemic exposure to DOX following the second meal of a dosing regimen of DOX or PARADOX medicated feeds distributed at 12-h intervals.
Test Herds with Porcine Respiratory Disease Complex.
This experiment took place in late spring, when temperature fluctuations and the risk of PRDC are highest (Stärk, 2000). We recruited 2 grower herds (herds X and Y) with recurrent PRDC of bacterial etiology, as diagnosed by clinical signs, necropsy lesions, and repeated isolation of pathogenic bacteria from the respiratory tract of diseased pigs. Pasteurella multocida, Haemophilus parasuis, and Streptococcus suis capsular types 2, 3, and 5 were recovered from both herds, and herd X also harbored Actinobacillus pleuropneumoniae and Bordetella bronchiseptica.
Herd X was in a farrow-to-finish operation, and herd Y was in a wean-to-finish operation. Both herds were housed in mechanically ventilated barns with all-in, all-out management per room. Pigs were kept in 15 m2 pens with concrete slatted floors, in groups of 12 to 13 pigs/pen (herd X) and of 13 to 15 pigs/pen (herd Y). As in most French grower barns, both herds were fed liquid diets served at 0800 and 1800. Troughs (5 m long, 
0.33 m/pig) were shared by 2 adjacent pens.
Medicated Feeds.
Batches of DOX and PARADOX dry medicated feeds (Porc croissance, Glon-Sanders; 9.44 MJ of NE, 16.7% CP) were delivered upon veterinary prescription in the 2 test herds (150 kg of each medicated article/herd). These were prepared using the same bags of drug premixes for the 2 herds (Doxyval 40 Porc P/M, Lot #2105098, potency = 41.3 mg/g; Concentrat V079 Paracétamol 100 Porc, Lot #2201793, potency = 100.0 mg/g), and were coded randomly feed A and feed B to ensure blind testing conditions. To compensate for age-related difference in medicated feed intake between the 2 farms (Labroue et al., 1995), nominal in-feed concentrations of PARA and DOX were set to 660 and 220 mg/kg, respectively, for herd X, and to 900 and 300 mg/kg, respectively, for herd Y.
Population PK Experiment and Inclusion Criteria.
The herds were visited 1 wk before the experiment to confirm the diagnosis of PRDC and to identify the rooms where respiratory signs were most prevalent. On the morning of experimental d 1, all pens in the preselected rooms were subject to medical inspection. First, the occurrence of dyspnea (labored breathing) and of tachypnea (>30 cycles/min) were recorded for each pig. In pens where the prevalence of these respiratory signs was >10%, rectal temperature was taken on all pigs to detect those with fever (i.e., >40.0°C rectal temperature). An animal was considered sick when it had fever and either tachypnea or dyspnea. Gender and other respiratory signs (cough, sneezing) were recorded individually, as well as housing descriptors (pen and trough identifications). Inclusion criteria for a pen were greater than 10% sick pigs, no more than 1 pig with an impaired ability to eat (e.g., lameness, wounds), and no antibiotics had been used in the pen for at least 15 d before the study.
In each farm, 8 eligible pens found within a single room were used for this study. All pigs from the included pens were weighed and ear-tagged. Pigs with feeding impairment (e.g., lameness) were removed from the pens for the duration of the study. No mixing of pigs from different pens was allowed. Pens were paired on the basis of prevalence of sick pigs and cumulative BW, and then feeds A and B were assigned randomly to each pen in the pair.
Because troughs were shared by 2 adjacent pens, feeds A and B were served in 2 troughs each, and then the pair-matched groups of pigs were placed in the pens corresponding to their treatment group. Whenever possible, groups of pigs stayed in their original pens.
To ensure that the sampling times would be respected, the medicated meals A and B were served in 2 troughs first, and in the 2 remaining troughs 30 min later. The amounts of feeds A and B dispensed in each trough cumulatively corresponded to 5 mg of DOX·kg of BW–1·meal–1. Feeds were mixed thoroughly with water at a ratio of 1:4 (vol/vol) immediately before the beginning of the meals. Meals were considered finished after 45 min of availability, after which feed surpluses were eliminated.
Pigs were restrained with a snare for vena cava blood puncture with vacuum tubes containing lithium heparin. Samplings were performed on a first-caught/first-bled fashion. The exact times of sampling after the beginning of the meal were recorded individually. Blood samples were centrifuged for 10 min at >2,200 x g and 4°C within 30 min of puncture. Plasma was immediately divided into 2 aliquots and stored at 20°C pending analysis.
Bioanalytical Methods.
For analysis of total plasma DOX concentration, DOX and the internal standard demeclocycline were extracted from plasma by precipitation with 70% perchloric acid. The HPLC apparatus consisted of a pump system fitted with an automatic injector and a variable-wavelength UV detector set to 355 nm. Separation was achieved by reverse phase column (Nucleosil C18, 3 µm, 150 x 4.6 mm, Interchim, Montluçon, France). The mobile phase was a mixture of acetonitrile:0.01 M oxalic acid (15:85, vol/vol) and was used at a flow rate of 0.6 mL/min. The within- and between-day CV were lower than 15%. The validated quantification limit of the assay was 0.05 mg/L.
For analysis of the total plasma concentrations of PARA and of its glucuronide metabolite (p-PARA), these molecules and the internal standard 3-acetamidophenol (3-PARA) were extracted from plasma by precipitation with 35% perchloric acid. The HPLC apparatus consisted of a pump system fitted with an automatic injector and an UV detector set at 254 nm. Separation was achieved by reverse phase column (Nucleosil C18, 5 µm, 125 x 4.0 mm) using a guard column (Ultrasep C18, 10 µm, 10 x 4.0 mm, Interchim, Montluçon, France). The mobile phase was a 95:5 (vol/vol) mixture of 0.01 M oxalic acid/methanol and was used at a flow rate of 0.3 mL/min. The within- and between-day CV of PARA and p-PARA were lower than 13%. The quantification limits of the assay were 0.1 mg/L for PARA and 1.0 mg/L for p-PARA.
Pharmacokinetic and Statistical Analyses.
Individual sets of plasma concentrations of DOX were subject to statistical moment PK analysis, where the AUC and AUMC of DOX were calculated with WinNonlin using the linear up/log down trapezoid rule. To take into account the actual dose of DOX offered in each trough, new variables were created for plasma concentrations of DOX and exposure parameters. For instance, the dose-normalized total concentration of DOX (Cp') was defined as:
 |
where Cp is raw DOX total concentration, 5 is the nominal dose of DOX, and Dose is the actual dose offered (both in mg/kg of BW). The following drug exposure parameters were obtained for all pigs: Cmax', AUC'0–11.5 (i.e., the partial AUC of Cp' measured during the first 11.5 h after the beginning of the second medicated meal) and 
Cp'0–11.5 (i.e., Cp' measured at 11.5 h, minus Cp' measured at –0.55 h).
Statistical exploratory data analysis (Ette et al., 2001) was performed with SAS (SAS Inst. Inc., Cary, NC) to identify the potential relationships between normalized DOX exposure parameters and plasma concentrations, along with physiological, pathological, therapeutic, and performance variables.
The potential relationships revealed with this exploratory analysis for Cp' were examined with the following linear repeated-measures mixed model using maximum likelihood estimation:
 |
where Yijklmnop is the response variable; µ is the grand mean; treatment (
i), herd (
j), gender (
k), disease (
l), and sampling time (
m) are fixed factors; the individual pig (Sn), trough (Fo), and pen (Pp) are random effects; the covariate pn is the body weight of the nth pig, with a being its coefficient; and 
ijklmnop is the random error term. Specific interactions involving treatment and disease were included in the model. Other interactions were dropped to avoid estimation problems. Differences between treatments were assessed separately at 4.5 h and at 11.5 h with specific a priori contrasts.
This model was estimated first with a free covariance structure. Then, the model was reestimated with more parsimonious covariance models (e.g., compound symmetry, autoregressive, Toeplitz), whose structures resembled that of the unstructured covariance matrix. The final selection of the covariance model was based on the Schwartz’s Bayesian criterion (Littell et al., 2000).
The effects of treatment, herd, gender, and disease on Cp'0–11.5 were addressed with a GLM using maximum likelihood estimation:
 |
In this equation, the response variable, fixed effects, and random error term are coded identically to the previous model. Likewise, the weight pm of the mth pig was a covariate of treatment.
Drug exposure parameters AUC'0–11.5, and Cmax', as well as the PK parameter tmax were submitted to separate stepwise forward regression analysis using DOX plasma concentration prior to the beginning of the second medicated meal, AUC of PARA, herd, treatment group, gender, body weight, and disease as predictor variables. We used a P = 0.15 marginal probability as threshold for inclusion or exclusion of a predictor variable in the model (Neter et al., 1990). Finally, within-treatment variability in AUC'0–11.5 was assessed with the Brown and Forsythe homogeneity of variance test on absolute deviation from group medians, which seems to be the best at providing power to detect variance differences while protecting the Type I error probability (SAS Inst. Inc., 1999).
|
RESULTS |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONIMPLICATIONSLITERATURE CITED |
|---|
The time-course of DOX concentrations in plasma after the in-feed and i.v. dosing are depicted in Figure 1, and its pharmacokinetic parameters are presented in Table 1. Measurable concentrations of DOX were detected for 24 h after in-feed dosing in 2 pigs, and the 24-h DOX concentration was just below the limit of quantification in 1 pig. During the absorption phase of the in-feed dosing, plasma concentrations of DOX increased at a variable rate and peaked between 3 and 5 h. The terminal elimination half-life of in-feed DOX was 4.63 ± 1.05 h (harmonic mean, jackknife estimate of the SD, as proposed by Lam et al., 1985).
|
|
Experiment 1b: Optimal Sparse-Sampling Strategy for DOX
Among the sets of sampling times that were generated from mathematical simulation, the one that gave minimal bias in AUC estimation had times starting at 1.38, 4.50, 6.17, and 11.50 h after giving the medicated meal. It was noteworthy that this set of sampling times included none of the times that, taken individually, produced the lowest mean square error in AUC estimation.
Experiment 2: Population Pharmacokinetics of DOX
Disclosure of the blinding code after completion of the study revealed that feeds A and B corresponded respectively to DOX and PARADOX. Table 2 presents the characteristics of the treated groups in each herd, and the doses offered on the evening and morning of experimental d 1 and 2. According to the actual in-feed concentrations of DOX, pigs in herd X were offered 113% of nominal dose, and pigs in herd Y were offered 96% of nominal dose. Pigs were no longer eating at the programmed end time of the meal, yet approximately 25% of the initial volume of feed was recovered from each trough.
|
. Drug concentrations reached a plateau between 1.83 and 6.17 h, and were declining by 11.50 h after the beginning of the second medicated meal. The interindividual variation in DOX plasma concentrations was large at each sampling time: the average CV of Cp' for the optimized times in the herds X and Y were, respectively, 35 and 43%, whereas those for the fixed times were, respectively, 70 and 84%. During the plateau phase, almost 60% of pigs in each herd had Cp' 0.5 mg/kg, of which about 27% had Cp' 0.75 mg/L. Approximately 5% of pigs in each herd had Cp' 
0.25 mg/L during this period of highest systemic exposure.
|
The repeated-measures mixed model revealed that Cp' was influenced mostly by sampling time (P < 0.001) and confirmed its dependence on BW: a 0.0021 ± 0.0009 mg/L decrease per kg of BW (P < 0.024). The covariance structure of data was approximated best with a first-order autoregressive model, and the individual pig was the largest source of variation in Cp'. Our model did not confirm the potential main effects on Cp' of gender, herd, and treatment (P > 0.15 in all cases), but suggested interactions involving disease. First, Cp' was subject to treatment x disease interaction (P = 0.099); healthy pigs given DOX had 0.06 mg/L lower Cp' than diseased ones (least squares means) but not pigs given PARADOX. In addition, a time x disease interaction on Cp' was detected (P = 0.14); healthy pigs had 0.05 mg/L higher concentrations at –0.55 h and 11.5 h.
Marginal t-tests revealed that Cp'0–11.5 was significantly greater than 0 for each herd and treatment group (P 0.002). The mixed-effect model revealed that this exposure indicator was affected by none of the tested variables and interactions (P > 0.36).
The distributions of AUC'0–11.5 and Cmax' as a function of herd and treatment are depicted in Figure 3. The exploratory analysis suggested that both estimators of exposure to DOX were affected by PARA, as well as gender in herd X. Both AUC'0–11.5 and Cmax' highly correlated with each other, but none of them correlated with BW or body temperature. Finally, tmax (i.e., the time of Cmax' occurrence) correlated negatively with BW (r = –0.22), but this was confirmed only for herd X when separate correlations were determined.
|
Table 3 presents the linear effects of independent variables on AUC'0–11.5 and Cmax'. As expected, plasma concentration of DOX prior to the beginning of the second medicated meal had major effect on the subsequent level of systemic exposure to this antibiotic. The stepwise forward procedure included a series of regressors related to PARA administration; PARA, the identifier of PARADOX pigs, had a negative coefficient for both exposure indicators. The other regressors were AUC0–11.5(PARA), the partial AUC of PARA measured between 0 and 11.5 h after the beginning of the meal, and Cpp-Para(t = 0), the concentration of p-PARA before beginning the second meal. In addition, the final regression model included body temperature and finishing, the identifier of herd Y. No other physiological, pathological, or performance variable had significant effect on the DOX exposure parameters.
|
|
DISCUSSION |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONIMPLICATIONSLITERATURE CITED |
|---|
), produces higher steady-state plasma concentrations in pigs (Pijpers et al., 1991a), and shows greater in vitro antimicrobial activity against porcine bacterial respiratory pathogens (Pijpers et al., 1989). Though most efficiently distributed through feed (McEwen and Fedorka-Cray, 2002), the clinical efficacy of in-feed antibiotics such as DOX likely is affected by feed intake.
Anorexia results from exposure to bacterial products (e.g., endotoxin, muramyl dipeptide), interleukins IL-1ß, IL-2, IL-6, and IL-8, tumor necrosis factor-
, and interferon. All these substances influence the central nervous regions that control energy and appetite homeostasis, indirectly (e.g., involving vagal afferents or leptin) or directly (Langhans, 2000). The reduction in-feed intake correlates with the concentration of PGE2 in cerebrospinal fluid (Lugarini et al., 2002), whose synthesis by intracerebral cyclooxygenase isoenzymes PARA selectively inhibits (Davies et al., 2004). In addition to its potent antipyretic and analgesic effects (Flower and Vane, 1972), PARA improves feed intake in laboratory animals exposed to endotoxin (Langhans et al., 1993), an effect that is partly unrelated to its other pharmacological effects (Langhans, 2000).
Prior to performing a farm-based study, an individual PK study was necessary because estimates of PK parameters of DOX in pigs are variable across studies (Pijpers et al., 1994; Anadón et al., 1996; Baert et al., 2000), no PK study has addressed the fate of DOX when coadministered with PARA, no PK study has used a commercial premix for feed administration of DOX or PARA, no PK study has addressed the PK of DOX when administered in liquid feed, and we needed optimal sampling times for accurately determining the AUC of DOX in a population of pigs.
Our open, 2-compartment model for simultaneous PK analysis of DOX plasma concentrations after in-feed and i.v. dosing provided excellent fit to observed data, especially during the first 12 h following each dosing. This model underestimated the through concentrations of DOX, mostly those of the i.v. administration, but goodness of fit with a 3-compartment mammillary model was not significantly better. Indeed, DOX concentrations observed between 48 and 54 h seem to parallel our predicted elimination slope, which suggests a nocturnal decrease in the elimination of DOX. We have reported a similar chronopharmacokinetic effect in swine for oxytetracycline and chlortetracycline (del Castillo, 2001). In dogs, a considerable amount of tetracycline leaves the urinary bladder and returns to the systemic bloodstream through lymphatic vessels (Milroy et al., 1974; Wood and Leonard, 1983). Hence, the normal urinary pH of pigs and their diurnal nature, which prolongs the retention of urine during the night, should favor the nocturnal cycling of urinary DOX.
The Vc and Vss of DOX estimated from our individual PK study was in the range of previously published results. However, our estimated CLt and CLd were higher than that of previous reports, and our elimination half-life was shorter than that reported by some authors. This discrepancy may be associated with differences in experimental settings; for instance, we have reported that the CLt of chlortetracycline and oxytetracycline increased with dietary Ca++ concentration (del Castillo, 2001). In addition, the CLt of DOX may decrease due to inadequate room temperature; chronic exposure to cold inhibits the expression of renal vasopressin receptors and leads to water loss and dehydration (Sun et al., 2003), a state that was reported to decrease significantly the CLt of oxytetracycline (Elsheikh et al., 1998). In our study, this source of bias should be minimal because our pigs were housed in a thermoneutral environment, i.e., 24.5°C (NRC, 1998). Finally, our estimated bioavailability of DOX was close to that of Baert et al. (2000), half of that reported by Sanders et al. (1996), and approximately 4 times lower than what was reported for human beings (Chambers, 2001). This may be associated with differences in DOX release from the oral dosing formulations (Baggot and Brown, 1998), with chyme composition (Nielsen and Gyrd-Hansen, 1996), intestinal mucus (Kararli, 1995), or P-glycoprotein expression (Kavallaris et al., 1993).
Mathematical simulations performed with the oral PK subsets of data gave place to an optimal sparse-sampling design that was different from the one in which each time is taken individually according to its performance, as measured by its ability in producing minimally biased results. This difference is a consequence of the purpose of this optimal sampling design, i.e., to obtain the most accurate individual estimates of drug exposure in the dosing interval. In this case, the shapes of individual PK curves need not be accurately determined as long as their associated AUC are minimally biased. If the design had been created by selecting the series of individually optimized sampling times, the biases in their associated trapezoids would accumulate instead of canceling each other (Wang, 2001). That is, our design is optimal for estimating systemic exposure to feed-administered DOX in swine but could not be used for another purpose.
In both herds where pigs were offered 5 mg of DOX/kg of BW in liquid feed at 12-h intervals, total DOX plasma concentrations partly overlapped the distributions of minimum inhibitory concentrations of bacteria involved in the PRDC (Bousquet et al., 1997). However, assuming that only free plasma concentrations are bacteriologically active and that the free DOX concentration in plasma of pigs is less than 10% of the total plasma concentration (Riond and Riviere 1990), it is likely that the current dosage regimen would be unable to achieve appropriate DOX exposure, i.e., plasma free DOX concentrations above the minimum inhibitory concentrations of the main pathogens in pigs. This aspect of our experiment will be fully addressed in another paper specifically devoted to the population PK/PD of DOX in pigs.
As illustrated in Figures 2
and 3, the interindividual variability in Cp' and systemic exposure to DOX was extremely large in both herds, both in DOX and PARADOX pigs. Approximately 40% of pigs had plasma concentrations of DOX below 0.5 mg/L during the period of maximum exposure, which is still comparable with the minimum inhibitory concentrations distributions of the most sensitive PRDC agents but considerably lower than those of A. pleuropneumoniae and P. multocida. This raises concerns about the risk of therapeutic failure and of selection of resistant bacteria in the least medicated pigs. To our knowledge, the dose-efficacy relationship of DOX against PRDC bacteria has not been reported, but those of oxytetracycline and chlortetracycline have been addressed with an A. pleuropneumoniae experimental challenge (Hunneman et al., 1994; del Castillo, 2001). Both antibiotics failed to control the disease in pigs whose plasma concentrations were about 25% the minimum inhibitory concentrations of the test pathogen. However, the antiinflammatory side effects of these drugs must be considered when assessing the therapeutic value of a dosing regimen. Many proinflammatory enzymes released by the phagocytes must bind to Zn++ for activation, a process that is impeded by the chelating ability of the tetracyclines (Golub et al., 1991).
Dietary PARA produced no significant improvement on Cp' and systemic exposure to DOX. Indeed, the recorded treatment x disease interaction on Cp' was restricted to DOX pigs; healthy pigs had lower Cp' than diseased ones. This finding and the time x disease interaction on Cp' agrees with the reported effects of A. pleuropneumoniae toxins on the PK of oxytetracycline (Pijpers et al., 1991b). Moreover, PARA may even increase the variability in systemic exposure, as indicated by the homogeneity of variance testing and the linear regression analysis; variance in Cmax' was largest in PARADOX pigs of herd X, and the regressor PARA had a negative coefficient, suggesting that PARA actually decreased the mean systemic exposure to DOX. Therefore, our hypothesis of a more even distribution of in-feed DOX among sick and healthy pigs due to the orexigenic effect of PARA receives little support from the results of our farm-based population PK study. Yet, these results do not disprove our hypothesis because AUC'0–11.5 and Cmax' are determined by CLt and F, both with large interindividual variation. The latter 2 PK parameters have not been measured for each pig in this study, but their contribution to variance in AUC'0–11.5 of DOX may be estimated to some extent. The PARA is a reliable tracer of gastric emptying (Willems et al., 2001), and its favorable PK properties makes it an indicator of feed intake. Hence, AUC0–11.5(PARA) gives insight to the variance components in the AUC'0–11.5 of DOX. Regression analysis for PARADOX pigs showed a linear increase in AUC'0–11.5 with AUC0–11.5(PARA), with an R2 = 0.25 when this regressor is used alone. This result reveals considerable variation in-feed intake among pigs and suggests that CLt and F were quite variable as well. The resulting interindividual differences in systemic exposure to DOX may preclude the detection of any orexigenic effect of in-feed PARA. In practice, the biopharmaceutic properties of the drug formulation and the social determinants of individual feed intake likely are the 2 most important factors in the level of systemic exposure to antibiotics in swine. Future research should focus first on these variables to maximize clinical efficacy and minimize public and environmental health risks associated with group antibiotic therapy in pigs.
Residual concentration of DOX prior to the second dosage (i.e., CpDoxy(t = 0)) was the most influential determinant in systemic exposure to DOX, as determined in the linear regressions of AUC'0–11.5 and Cmax' of DOX. These results were expected because our dosing regimen allowed DOX to accumulate, as shown by a positive Cp'0–11.5. Another variable related to feed intake is Finishing, the identifier of herd Y. This variable, which had positive effect on Cmax', may describe the effect of body size on the rate of feed intake (del Castillo et al., 2002). Older pigs may eat the antibiotic dose in a shorter time, which would increase the rate of drug absorption and Cmax. This is further supported by the negative correlation value of body weight on tmax. Therefore, our results show the importance of understanding the determinants of appetite and of the fate of feed-administered drugs for a judicious use of antibiotics in swine herds.
A reason for the negative effect of PARA on AUC'0–11.5 and Cmax' of DOX may be its deterrent effect on feed intake. The bitterness of PARA is associated with noncompliance in children, which has prompted the design of more palatable dosing forms to overcome this problem (Suzuki et al., 2004). A similar effect might be observed in pigs, which possess high taste acuity for bitter substances (Danilova et al., 1999). Hence, this result suggests that the enhanced clinical efficacy of PARADOX medicated feeds is related to the antipyretic or analgesic effects, or both, of PARA, not to its orexigenic effect.
Overall, provided that DOX is prescribed at an adequate dose for the therapeutic drug regimen and that the feed intake of treated pigs is not abolished, the systemic exposure to DOX in sick pigs may be sufficient to sustain its therapeutic efficacy.
|
IMPLICATIONS |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONIMPLICATIONSLITERATURE CITED |
|---|
|
Footnotes |
|---|
2 Current address: Université de Montréal, Faculté de médecine vétérinaire. CP 5000 St-Hyacinthe (Qc), J2S 7C6 Canada.
3 Corresponding author: pl.toutain{at}envt.fr
Received for publication September 29, 2005. Accepted for publication May 16, 2006.
|
LITERATURE CITED |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONIMPLICATIONSLITERATURE CITED |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME |
