Journal of Antimicrobial Chemotherapy (2001) 48, 813-820
© 2001 The British Society for Antimicrobial Chemotherapy
A pharmacokinetic/pharmacodynamic approach to show that not all fluoroquinolones exhibit similar sensitivity toward the proconvulsant effect of biphenyl acetic acid in rats
Equipe ‘Médicament et BHE’, Laboratoire de Pharmacie Galénique et de Biopharmacie, Faculté de Médecine and Pharmacie, 34 rue du Jardin des Plantes, BP 199, 86005 Poitiers Cedex, France
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
The proconvulsant effect of biphenyl acetic acid (BPAA) on several fluoroquinolones (FQs) was investigated in vivo, by measuring drug concentrations in the biophase at the onset of convulsions. Male Sprague–Dawley rats (n = 134) were given BPAA orally, at various doses 1 h before starting FQ infusion, which was maintained until the onset of maximal seizures, when cerebrospinal fluid (CSF) and plasma samples were collected for drug concentration determination. The FQ–BPAA interactions in the biophase (CSF) were adequately described on most occasions by an inhibitory Emax effect model with a baseline effect parameter. The efficacy of the proconvulsant effect was characterized by the ratio of the CSF concentrations of FQs at the onset of convulsant activity when BPAA was absent (CCSF0, FQs) and as BPAA CSF concentrations tended toward infinity (CCSFbase, FQs). This ratio varied from 15 for enoxacin to 1.9 for sparfloxacin. The potency of the proconvulsant effect was characterized by the CSF concentration of BPAA corresponding to a proconvulsant effect half of its maximum. This parameter varied between 0.18 ± 0.06 µmol/L with enoxacin and 15.0 ± 12.1 µmol/L with sparfloxacin. The CSF diffusion of all FQs was apparently non-linear, as well as the plasma protein binding of BPAA, complicating interpretation of plasma data. The important variability in the proconvulsant effect of BPAA demonstrated in this study between various FQs suggests that in vitro
-aminobutyric acid (GABA) binding experiments conducted in the presence of BPAA are unlikely to be good predictors of FQ convulsant risk in clinical practice. |
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Fluoroquinolones (FQs) represent an important class of antimicrobial agents and have shown efficacy in the treatment of various bacterial infections.1,2 Although FQs are associated with a low incidence of central nervous system (CNS) disorders including headache, confusion, dizziness, anxiety, nervousness and nightmares, they may occasionally induce convulsive seizures,3,4 especially in patients with impaired hepatic or renal function, or in patients receiving FQs in combination with non-steroidal anti-inflammatory drugs (NSAIDs) such as fenbufen.3 Although it is generally assumed that FQ convulsant activity is related to
-aminobutyric acid (GABA) inhibition, these antibiotics, with the exception of norfloxacin, have a weak affinity for GABAA receptors, which in fact is too low to be determined accurately.5 However, this affinity is dramatically increased in the presence of biphenyl acetic acid (BPAA), the active metabolite of fenbufen.5 For that reason most in vitro experiments including electrophysiological6–11 and radioligand binding5,12 studies have been conducted in the presence of BPAA. It is questionable whether such experiments help in predicting the convulsant risk associated with FQ administration in the usual clinical situation, that is without concomitant use of NSAIDs. The answer is probably no, for at least two main reasons. One is that in vitro experiments do not take into account the CNS diffusion characteristics of FQs, which vary considerably between compounds.13,14 The other reason is that the proconvulsant effect of BPAA may also vary between FQs in terms of both potency and efficacy, which to our knowledge has never been assessed with appropriate methodologies. We have recently developed a modelling approach to characterize the proconvulsant effect of BPAA on norfloxacin based upon determination of drug concentrations in CSF, shown previously to be part of the biophase, at the onset of activity.15 This approach has been used in the present study to demonstrate that the proconvulsant effect of BPAA is highly variable among FQs and therefore that the in vitro GABA binding experiments in the presence of BPAA are not appropriate to predict the convulsion risk associated with FQs in clinical practice. |
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Animals
This work was carried out in accordance with Principles of Laboratory Animal Care (NIH Publication #85-23, revised 1985), and the study protocol was approved by the local ethics committee. Male Sprague–Dawley rats (n = 134) from Depres Breeding Laboratories (St Doulchard, France), were housed in the animal breeding facilities of the laboratory. Their body weights ranged from 215 to 280 g with an average of 243 ± 14 g (mean ± S.D.). Animals were placed in wire cages in a 12 h light–dark cycle for 5 days before the beginning of the experiment. During this period, they had free access to food (UAR A04; UAR Laboratories, Villemoisson-sur-Oge, France) and water.
Implantation of cannula
The day before the experiment, animals were anaesthetized with an ip injection of 60 mg/kg sodium pentobarbital (Sanofi Laboratories, Libourne, France). A polyethylene cannula [0.58 mm internal diameter (i.d.), 0.96 mm outside diameter; Harvard, Les Ulis, France] was then implanted in the right jugular vein. Following surgery, the rats were kept under a heating lamp, and after the first signs of movement, the animals were placed into individual plastic cages. Food was withdrawn 12 h before the experiment, but the animals had free access to water until drug infusion.
Drug administration
The day after surgery, a BPAA suspension was given orally by gastric tube 1 h before the beginning of the FQ infusion. The BPAA suspension was prepared for different doses (depending on the FQ infused) by a mixture of BPAA (Sigma, St Quentin Fallavier, France) in sodium carboxymethyl cellulose (French Pharmaceutics Cooperation, Melun, France) 0.5% (w/v) (Table 1
). A solution of sodium carboxymethyl cellulose 0.5% devoid of BPAA was administered orally for control groups (n = 5–9). FQs were administered as 240 mM solutions, infused through the jugular cannula connected to a motor-driven syringe pump (Program 2; Vial Medical, Brezins, France) at a flow rate of 4 mL/h (960 µmol/h). A commercial solution of pefloxacin methane sulfonate (80 mg/mL or 240 mmol/L) (Roger Bellon Laboratories, Neuilly sur Seine, France) was used for administration. Enoxacin, fleroxacin and sparfloxacin were dissolved in a minimal volume of 1 M NaOH to which an equal volume of phosphate buffer (pH 7.4) was added, the pH was subsequently adjusted to 11.0 with 1 M HCl, and the final concentration was adjusted to 240 mM by addition of a 5% glucose solution, as described previously.14 At the onset of maximal seizures, the FQ infusion was stopped. Drug administrations were conducted between 2:00 p.m. and 7:00 p.m.
|
Sample collection
Immediately after exhibiting maximal seizures, rats were anaesthetized with an im injection of a mixture of two-thirds ketamine (Ketalar; Parke-Davis, Courbevoie, France) and one-third xylazine hydrochloride (Rompun 2%; Bayer Laboratories, Puteaux, France), unless they had died following maximal seizures. In any case, CSF and blood were collected within 3 min, as described previously.16 Blood was immediately centrifuged at 1000g for 10 min (GR 412 model; Jouan, St Herblain, France). Plasma was transferred into two separate tubes. One fraction was kept frozen at -20°C until assayed. The other fraction was ultrafiltered with a Centrifree system (CF50A model; Amicon, Molshein, France) for determination of unbound concentrations.
Drug analysis
FQs and BPAA concentrations were determined by high-performance liquid chromatography (HPLC). The chromatographic system consisted of a Shimadzu LC-6A pump connected to a Waters 717 plus autosampler and to a Waters 470 fluorimetric detector. Excitation and emission wavelengths were, respectively, 280 and 445 nm for pefloxacin, 287 and 440 nm for fleroxacin, and 268 and 400 nm for enoxacin. A Waters 484 tunable absorbance detector was used for UV detection at 364 nm and 254 nm for sparfloxacin and BPAA, respectively. Data were recorded and processed using a Waters 746 integrator.
FQ concentrations in CSF, ultrafiltrate (UF) and plasma were determined using a methodology described previously.14 The mobile phase consisted of 0.1 M aqueous citric acid solution containing 7–13% (v/v) acetonitrile and 10 mM tetra-butyl ammonium perchlorate, and the flow rate was 0.8 mL/min. FQs were assayed in CSF and UF by direct injection after appropriate dilution in 0.1 M citrate buffer. Plasma samples were diluted appropriately by addition of 1.7% (v/v) perchloric acid. The mixture was then centrifuged (1000g, 10 min, 5°C) and the supernatant was injected on to the column (Kromasil C18; 5 µm, 150 x 3 mm i.d.).
For BPAA determination, the mobile phase consisted of a mixture of 50% acetonitrile, 49% water (Milli-Q) and 1% acetic acid (by vol), and the flow rate was 0.8 mL/min.15 BPAA was assayed in CSF and UF by direct injection after appropriate dilution in NaCl 0.9%. Plasma samples were diluted appropriately by addition of 1.7% (v/v) perchloric acid. The mixture was then centrifuged (1000g, 10 min, 5°C) and the supernatant was injected on to the column (Kromasil C18 column, 5 µm, 150 x 3 mm i.d.).
Data modelling
Modelling of FQ–BPAA interaction in CSF..
CSF FQ versus BPAA concentrations were fitted according to an inhibitory Emax effect model with a baseline effect parameter15 with a general form as follows (Equation 1):
 | (Equation 1) |
 |
CSF diffusion modelling.. Two different models were fitted to the CSF versus UF concentrations of both FQs and BPAA:
 | (Equation 2) |
 | (Equation 3) |
Plasma protein binding modelling..
FQ unbound versus plasma total concentrations were linearly related and fitted according to the following equation.
 | (Equation 4) |
BPAA bound (CB) versus free plasma (CU) concentrations were fitted according to the Langmuir equation (Equation 5).17,18
 | (Equation 5) |
Modelling was performed with WinNonlin, version 1.1 (SCI software, Carry, NC, USA), with uniform weighting. Discrimination between linear and non-linear models of CSF diffusion and protein binding was assessed from different criteria, including visual inspection, residual analysis, sum of squared residuals (SSR), correlation coefficient between observed and predicted values, and Akaïke information criteria (AIC).
Statistics
Results are presented as mean ± S.D. FQ doses and time necessary to obtain maximal seizures in the presence or absence of BPAA were compared by the non-parametric Kruskal–Wallis test followed by Dunn's multiple comparison post test (GraphPad PRISM software). Results obtained in this study with enoxacin, sparfloxacin, pefloxacin and fleroxacin are presented together with those published previously for norfloxacin,15 to facilitate comparisons.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
In the absence of BPAA, maximal seizures appeared at various times of infusion, ranging from 21.4 ± 4.0 min with sparfloxacin to 41.7 ± 3.9 min with enoxacin (Table 1
), and corresponding to convulsant doses from 1394 ± 247 µmol/kg (sparfloxacin) to 2775 ± 388 µmol/kg (enoxacin) (Figure 1), consistent with previously published values.14 The reduction of these convulsant doses in the presence of BPAA was also highly variable from one FQ to another. The highest reduction, both in absolute and relative terms, was observed with enoxacin, whereas BPAA had apparently virtually no proconvulsant effect on fleroxacin (Figure 1).
|
At the CSF concentration level, the inhibitory Emax effect model with a baseline effect parameter (Equation 1
) adequately described the enoxacin, sparfloxacin and pefloxacin data (Figure 2) with parameter estimates presented in Table 2, but not the fleroxacin data. Apparently, fleroxacin CSF concentrations were linearly related to BPAA CSF concentrations, with an intercept (CCSF0) and a slope value, respectively, equal to 237 µM and -0.239.
|
|
The relationship between CCSF and CU was non-linear (Equation 2
) with enoxacin, sparfloxacin and pefloxacin, as observed previously with norfloxacin.15 Again, modelling of fleroxacin data raised problems since none of the two potential models (Equations 2 and 3) applied (Figure 3). CCSF and CU of BPAA were virtually identical, as observed previously (data not shown).15
|
Finally, the protein binding of these FQs was linear (Equation 4
) and relatively limited, with unbound fractions (fu) estimated by linear regression analysis to be 0.61 ± 0.05 for sparfloxacin, 0.64 ± 0.01 for fleroxacin, 0.80 ± 0.02 for pefloxacin and 0.83 ± 0.01 for enoxacin. Very similar fu values were obtained by averaging the individual CU/CP ratios of each FQ, in agreement with results published previously.14 Conversely, the plasma protein binding of BPAA was extensive and non-linear, again as described previously,15 but estimates of the parameters characteristic of the Langmuir equation (Equation 5), in particular CBmax, were apparently lower in the presence of norfloxacin and enoxacin than in the presence of other FQs (Figure 4).
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
The convulsant doses of FQ administered iv were most often reduced significantly after pretreatment with sufficiently high oral doses of BPAA. A lot of variability was observed between FQs, but due to the lack of knowledge of the oral bioavailability of BPAA, precise data interpretation is impossible at the dose level. Analysis of plasma concentrations would help to solve this problem, except that non-linear protein binding and/or non-linear distribution within the biophase would complicate data interpretation.15
FQ plasma protein binding was limited and linear but that of BPAA was extensive and non-linear. However, CBmax estimates were apparently lower in the presence of norfloxacin and enoxacin than in the presence of other FQs (Figure 4). This observation was surprising because FQs are unlikely to alter the protein binding of BPAA. It is most likely due to the fact that unbound plasma concentrations of BPAA at the onset of convulsant activity were much lower in the presence of norfloxacin and enoxacin, due to the high convulsant effect of these mixtures, than in the presence of sparfloxacin, pefloxacin or fleroxacin. As a consequence, although the Langmuir equation applied to norfloxacin and enoxacin data, corresponding CBmax estimates were probably not accurate.
The CSF diffusion of all FQs except fleroxacin was also apparently non-linear, with CCSF increasing less than proportionally to CU (Figure 3), meaning that lower CCSF/CU ratios were observed at higher FQ concentrations, that is at lower BPAA levels. However, not only FQ concentrations, but also infusion duration, were dramatically increased as BPAA doses and concentrations decreased (Table 1). Owing to their slow CSF diffusion, changing the infusion duration of FQs may affect the CCSF/CU ratio at the onset of activity, at least for the most hydrophilic FQs.16
Non-linearities in the BPAA plasma protein binding and CSF diffusion of FQs identified during this study indicate that the proconvulsant activity of BPAA on FQs should be interpreted cautiously from plasma concentration measurements. However, these non-linearities cannot be well characterized because the present study was not designed to do so, but rather to compare the proconvulsant effect of BPAA on various FQs from drug concentration measurements within the biophase (CSF) at the onset of convulsant activity. Within this fluid, the inhibitory Emax model with baseline effect, used previously to describe the proconvulsant effect of BPAA on norfloxacin, applied to other FQs, with the exception of fleroxacin. The CSF concentrations of this particular FQ at the onset of activity was virtually unaffected by the presence of BPAA, even at CSF concentrations much higher than the CCSF50,BPAA values estimated for other compounds (
100 µmol/L) (Figure 2). This confirms observations made at the dose level (Figure 1 and Table 1). Visual inspection of the fleroxacin versus BPAA CSF concentrations at the onset of activity (Figure 2) suggested the existence of a linear relationship between the two, which was confirmed by regression analysis, with a slope value K estimated as -0.239. In fact this linear relationship can be considered as a reduced form of the inhibitory Emax model with baseline effect when CCSF, BPAA << CCSF50, BPAA, the slope factor K then being equal to -(CCSF0, FQs - CCSFbase, FQs)/CCSF50, BPAA.
With other FQs, CCSF0, FQs and CCSFbase, FQs were reasonably well estimated possibly with the exception of the asymptotic value of enoxacin concentration when that of BPAA tended toward infinity (Table 2). The maximal proconvulsant effect of BPAA may then be assessed and compared between FQs from the corresponding CCSF0, FQs/CCSFbase, FQs ratios, which characterize the efficacy of the proconvulsant effect, whereas the CSF BPAA concentrations at which half of the proconvulsant effect is observed (CCSF50, BPAA) characterize the potency of the effect. The efficacy of the proconvulsant effect is highly variable among the five FQs tested, since the CCSF0, FQs/CCSFbase, FQs ratios vary from 15 for enoxacin to 1.9 for sparfloxacin and even less for fleroxacin. The potency is also highly variable from one FQ to another since the CCSF50, BPAA values vary from 0.18 ± 0.06 µmol/L for enoxacin to 15.0 ± 12.1 µmol/L for sparfloxacin and even more for fleroxacin. Interestingly, a similar ranking appears between FQs in terms of potency and efficacy (Table 2).
Because of the important variability in the proconvulsant activity of BPAA among FQs, the in vitro GABA binding experiments in the presence of BPAA traditionally used to predict the convulsant potential of FQs are likely to be meaningless. In order to characterize the affinities of FQs for GABAA receptors, we previously estimated IC50 values in the presence of BPAA, to be of the order of 10–8 mol/L for norfloxacin and enoxacin, and 10–4 mol/L for sparfloxacin, fleroxacin and pefloxacin,14 suggesting that the convulsant potential of enoxacin was much higher than that of sparfloxacin. However, during the same study we also found that the CSF concentrations of these two FQs at the onset of convulsant activity in rats were almost identical (respectively, 150 ± 19 µmol/L and 131 ± 27 µmol/L), suggesting that their convulsant potentials were actually very similar. These conflicting data can probably be explained by, or at least are consistent with, the much higher proconvulsant effect of BPAA demonstrated on enoxacin than on sparfloxacin in the present study. It would obviously be interesting to compare the proconvulsant activity of BPAA in the presence of various FQs determined in vivo in this study, with its effect on the GABAA affinity of these FQs in vitro; however, this is not possible, because with the exception of norfloxacin, the affinity of FQs for GABA receptors in the absence of BPAA is too low to be determined accurately.5
In conclusion, this study provides indirect evidence that the traditional in vitro GABA binding experiments in the presence of BPAA are not adequate to predict the convulsion risk associated with FQs, not only because as in vitro investigations they do not capture the variable ability of compounds to reach receptors at the central level that is a determinant predictor of FQ convulsant activity in vivo,14 but also because of the presence of BPAA, which has an important proconvulsant effect on certain FQs such as enoxacin, but virtually no effect on others such as fleroxacin.
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
The technical collaboration of Isabelle Martineau was greatly appreciated during this study.
|
Notes |
|---|
* Corresponding author. Tel: +33-5-49-45-43-79; Fax: +33-5-49-45-43-78; E-mail: william.couet{at}univ-poitiers.fr
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
1 . Moellering, R. C. (1996). The place of quinolones in every clinical practice. Chemotherapy 42, 54–61.
2 . Hooper, D. C. (1998). Clinical applications of quinolones. Biochimica et Biophysica Acta 1400, 45–61.[Medline]
3 . Christ, W. (1990). Central nervous system toxicity of quinolones: human and animal findings. Journal of Antimicrobial Chemotherapy 26, 219–25.
4 . Hori, S. & Shimada, J. (1993). Effects of quinolones on the central nervous system. In Quinolones Antimicrobial Agents, 2nd edn, (Hooper, D. C. & Wolfon, J. S., Eds), pp. 513–8. American Society for Microbiology, Washington, DC.
5
.
Akahane, K., Sekiguchi, M., Une, T. & Osada, Y. (1989). Structure–epileptogenicity relationship of quinolones with special reference to their interaction with -aminobutyric acid receptor sites. Antimicrobial Agents and Chemotherapy 33, 1704–8.
6
.
Akaike, N., Shirasaki, T. & Yakushiji, T. (1991). Quinolones and fenbufen interact with GABAA receptors in dissociated hippocampal cells of the rat. Journal of Neurophysiology 66, 497–504.
7
.
Halliwell, R. F., Davey, P. G. & Lambert, J. J. (1991). The effect of quinolones and NSAIDs upon GABA-evoked currents recorded from dorsal root ganglion neurones. Journal of Antimicrobial Chemotherapy 27, 209–18.
8 . Halliwell, R. F., Davey, P. G. & Lambert, J. J. (1995). A patch clamp study of the effects of ciprofloxacin and biphenyl acetic acid on rat hippocampal neurone GABAA and ionotropic glutamate receptors. Neuropharmacology 34, 1615–24.[Web of Science][Medline]
9 . Yakushiji, T., Shirasaki, T. & Akaike, N. (1992). Non-competitive inhibition of GABA responses by a new class of quinolones and non-steroidal anti-inflammatories in dissociated frog sensory neurones. British Journal of Pharmacology 105, 13–8.[Web of Science][Medline]
10 . Green, M. A. & Halliwell, R. F. (1997). Selective antagonism of the GABAA receptor by ciprofloxacin and biphenylacetic acid. British Journal of Pharmacology 122, 584–90.[Web of Science][Medline]
11 . Kawakami, J., Yamamoto, K., Asanuma, A., Yanagisawa, K., Sawada, Y. & Iga, T. (1997). Inhibitory effect of new quinolones on GABAA receptor-mediated response and its potentialisation with felbinac in Xenopus oocytes injected with mouse brain mRNA: correlation with convulsive potency in vivo. Toxicology and Applied Pharmacology 145, 245–54.
12
.
Akahane, K., Kimura, Y., Tsutomi, Y. & Hayakawa, I. (1994). Possible intermolecular interaction between quinolones and biphenylacetic acid inhibits -aminobutyric acid receptor sites. Antimicrobial Agents and Chemotherapy 38, 2323–9.
13
.
Jaehde, U., Langemeijer, M. W. E., de Boer, A. G. & Breimer, D. D. (1992). Cerebrospinal fluid transport and disposition of the quinolones ciprofloxacin and pefloxacin in rats. Journal of Pharmacology and Experimental Therapeutics 263, 1140–6.
14
.
Delon, A., Bouquet, S., Huguet, F., Brunet, V., Courtois, Ph. & Couet, W. (1999). Pharmacokinetic–pharmacodynamic contributions to the convulsant activity of fluoroquinolones in rats. Antimicrobial Agents and Chemotherapy 43, 1511–5.
15 . Marchand, S., Pariat, C., Bouquet, S., Courtois, Ph. & Couet, W. (2000). Pharmacokinetic–pharmacodynamic modelling of the convulsant interaction between norfloxacin and biphenyl acetic acid in rats. British Journal of Pharmacology 129, 1609–16.[Web of Science][Medline]
16
.
Delon, A., Huguet, F., Courtois, Ph., Vierfond, J. M., Bouquet, S. & Couet, W. (1997). Pharmacokinetic–pharmacodynamic contributions to the convulsant activity of pefloxacin and norfloxacin in rats. Journal of Pharmacology and Experimental Therapeutics 280, 983–7.
17 . Benincosa, L. J. & Morris, M. E. (1993). Nonlinear pharmacokinetic and protein binding of tiaprofenic acid in female Lewis rats. Journal of Pharmaceutical Sciences 82, 429–30.[Web of Science][Medline]
18 . Kochak, G. M., Sudhakar, P., Robert, I., Honc, F., Kachman, D., Perrino, P. & Egger, H. (1993). Prinomide tromethamide pharmacokinetics: Mutally dependent saturable and competitive protein binding between primonide and its own metabolite. Pharmaceutical Research 10, 49–55.[Web of Science][Medline]
Received 2 February 2001; returned 4 July 2001; revised 31 July 2001; accepted 3 September 2001
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||