Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

The current knowledge about the mechanism of action of disease-modifying antirheumatic drugs (DMARDs) involves inhibition of the production of proinflammatory cytokines (Bondeson, 1997). In particular, tumor necrosis factor- (TNF-) has been demonstrated to be of pivotal importance in rheumatoid arthritis (RA). The pharmacological effect of anti-inflammatory and immunomodulating agents is commonly studied with bacterial lipopolysaccharide (LPS) (Remick et al., 1989; Shapira et al., 1996) because administration of LPS into experimental animals leads to a rapid induction of macrophage-monocyte TNF- synthesis and develops similar pathophysiological changes to those of an inflammatory response. Hence, the effect of DMARDs on LPS-induced elevated TNF- exposure serves as a useful pharmacodynamic (PD) model for these types of agents.

In this study, we investigated the pharmacological action of susalimod, a metabolically stable chemical analog of sulfasalazine designed within a drug development program for the treatment of RA. The pharmacokinetics (PK) of susalimod has been investigated in various animal species (Påhlman et al., 1998). Its PK profile is characterized by an extensive biliary excretion, mainly as unchanged parent drug. Furthermore, the compound is very highly bound to plasma albumin and has a small volume of distribution. In contrast to previously reported studies, which to our knowledge have only sought to describe a dose-response relationship, we have tried to establish a concentration-effect relationship between a compound (susalimod) and cytokine serum levels.

Integrated PK-PD modeling deals with the issue of correlating the time course of pharmacological effect intensity to the plasma pharmacokinetics of a drug. With its potential use in comparison of in vivo potency and in prediction of outcomes, PK-PD modeling is recognized to be of critical importance in the knowledge-gathering process of drug development (Yacobi et al., 1993; Breimer and Danhof, 1997). The possibility to perform relevant PK-PD modeling (i.e., in which adequate response variables are chosen and the rate-limiting steps are accurately identified) is very much dependent on the knowledge about the molecular biology of drug action. In the common case where the pharmacological mechanism of action has not been fully clarified, modeling may nevertheless be viewed as a useful tool in exploring these events (Sheiner et al., 1979; Holford and Sheiner, 1981; Boxtel et al., 1992; Dayneka et al., 1993; Jusko and Ko, 1994; Danhof and Mandema, 1995).

In this study, serum concentration-time data of the immunomodulating agent susalimod and cytokine TNF- were explored following concomitant administration of LPS and four different doses of susalimod to mice. In addition to TNF-, other cytokines that are known to play a role in the biological response to LPS, e.g., the proinflammatory cytokine interleukin (IL)-6 as well as the anti-inflammatory cytokine IL-10, were determined. The effect of susalimod on these cytokines was found to be most evident for TNF-. The concentration-effect relationship between susalimod and TNF- was described by a specially designed integrated PK-PD model, comprising a three-compartment PK model and a mechanistically based indirect PD model.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Chemicals

Bacterial LPS (Escherichia coli 0127:b8), batch 10692JE, was provided by Difco (Detroit, MI). LPS was used as a solution in 0.9% NaCl; 50 µg/ml. Susalimod, 2-hydroxy-5-[[4-[(3-methyl-2-pyridinylamino)sulfonyl]-phenyl]ethynyl]benzoic acid, mol wt 408 g/mol, was provided by Pharmacia & Upjohn (Uppsala, Sweden). Susalimod was used as a solution in 0.9% NaCl (adjusted to pH 8.0) at concentrations of 1.25, 2.5, 5, or 10 mg/ml. Other chemicals used (see below) were of analytical grade and were obtained commercially.

Experimental Design

Two hundred female CD-1 mice, provided by Charles River, Germany, were used. The experiment was approved (C330/95) by the Animal Ethics Committee of Uppsala, Sweden. The mice were ~10 weeks old and weighed 20 to 25 g. A sublethal endotoxin shock was induced in all mice by administration of LPS, 1 mg/kg. Four groups received susalimod (25, 50, 100, or 200 mg/kg) in connection with the LPS administration, and one control group was given saline to explore the induced cytokine exposure in the absence of susalimod. The test articles were administered i.p., 0.02 ml/g b.wt., and blood samples were drawn from orbital plexus under anesthesia (Metofane; Mallinckrodt Inc., St. Louis, MO) at 15, 30, 45, 60, 90, 120, 180, and 300 min after administration. Blood was collected from five mice at each time point. Each animal was sampled once and then sacrificed by neck dislocation. Serum was prepared and analyzed with respect to susalimod and TNF-, IL-6, and IL-10.

Analyses

Susalimod was determined in serum by reversed-phase liquid chromatography and UV-absorbance detection. The serum samples were mixed in equal volumes with distilled water or phosphate buffer. The protein precipitation was performed with methanol by using the volume ratio 3:2 for methanol and sample, respectively. After centrifugation, the supernatant was diluted 1:1 with distilled water, and an aliquot (20 µl) of the solution was injected into the HPLC system (Varian 9010, Varian, Palo Alto, CA; LDC SpectroMonitor 4100, Thermoquest, San Jose, CA; and Laboratory Data System Nelson 2600, Perkin-Elmer, Norwalk, CT). Single determination of susalimod was performed for the study samples. Separation was performed on a Kromasil C₈ column (5 µm, 150 mm × 4.6 mm i.d.) with a guard column Kromasil C₈ (5 µm, 10 mm × 3.2 mm i.d.). The mobile phase was methanol/phosphate buffer (0.05 M, pH 7.0), with a methanol range of 50 to 60%. Quantification was performed with UV-absorbance detection at 320 nm with an external standard. The standard curve in serum was linear in the range from 0.62 to 150 µM susalimod. The accuracy and precision of the method were continuously examined during the study by analyzing duplicates of spiked quality control samples at concentrations of 0.62, 74, and 136 µM susalimod. The interassay accuracy for susalimod varied between 100 and 103%. The interassay variation (expressed as relative S.D.) was 14% at 0.62 µM (low limit of quantification of susalimod) and <2% at 74 and 136 µM susalimod.

The serum cytokine concentrations were analyzed with highly specific sandwich enzyme-linked immunosorbent assays (ELISAs). ELISA protocols supplied by PharMingen (San Diego, CA) were used with a few minor modifications. All incubations were performed on flat-bottom high-binding polystyrene microtiter plates (Nunc Immonoplate Maxisorp F-96; A/S Nunc, Roskilde, Denmark), with specific rat anti-mouse monoclonal antibodies against TNF- (catalog no. 18121D), IL-10 (catatalog no. 18141D), and IL-6 (catalog no. 18071D). Recombinant mouse TNF-, IL-10, and IL-6 proteins were used as standards. For detection of cytokines bound to the primary capture antibodies, biotinylate rat anti-mouse TNF-, IL-10, and IL-6 antibodies were used. All antibodies (primary and secondary), as well as recombinant cytokines, were purchased from PharMingen (San Diego, CA). Visualization of the amount of cytokines in each well was performed by the addition of avidine complexed with biotinylated alkaline phosphatase (ABComplex/AP, DAKOPATTS, Älvsjö, Sweden) and the phosphatase chromogene p-nitrophenylphosphate (1 mg/ml; Sigma Chemical Co., St. Louis, MO) in diethanolamine buffer, pH 9.8. After incubation in darkness, absorbance (E_max) was measured at 405 nm in a microplate reader (Molecular Devices Corp., Menlo Park, CA). All measurements (standards, samples) were performed as duplicates. The detection limit was defined as the blank absorbance + 3 S.D. Detection limits of the TNF-, IL-10, and IL-6 assay were estimated to be 0.010, 0.065, and 0.70 ng/ml, respectively.

PK-PD Modeling

PK Model. The absorption, distribution, metabolism, and excretion characteristics of susalimod have been thoroughly investigated in various animal species, especially rat, dog, and monkey, but also mouse, rabbit, and mini-pig (Påhlman et al., 1998). In these studies, it has been shown that susalimod is mainly cleared via the bile without prior biotransformation, that clearance decreases with increasing dose, that plasma protein binding is very high, and that its volume of distribution is small. Furthermore, it has been shown in rats that enterohepatic recirculation occurs. Plasma PK data obtained in previous studies indicate that enterohepatic recirculation of susalimod occurs also in mouse and other animal species (unpublished data).

View larger version (16K): [in this window] [in a new window]   Fig. 1.   Integrated PK-PD model, including a three-compartment PK model of susalimod and a PD model of TNF- after LPS administration to female CD-1 mice.

(1)

(2)

(3)

PD Model. The induction of TNF- synthesis following LPS administration was described by an indirect PD model, including an intracellular compartment and a serum compartment. It was assumed that LPS induced the TNF- synthesis during a limited time interval, t_synt. The synthesis rate was described by a zero-order rate constant, k₀, whereas the release rate (or appearence rate in serum), k_in, of the synthesized amount of TNF- as well as the disappearance rate from serum, k_out, were assumed to follow first-order kinetics. The distinct time lag between LPS administration and TNF- appearance in serum was modeled by the parameter t_lag.

(4)

(5)

PK-PD Model. The relationship between the serum concentration of susalimod and the induced TNF- serum levels was described by an indirect-response PK-PD model (Fig. 1). In the integrated model, the synthesis rate was multiplied by an inhibition function, I(t), governed by the serum concentration time course of susalimod:

(6)

The maximum inhibitory effect, I_max, was set to 1, i.e., the synthesis of TNF- was assumed to be fully inhibited at high enough susalimod levels. IC₅₀ denotes the potency of susalimod, i.e., the serum concentration that produces 50% of I_max. A slope factor, n, also was included in the Hill equation.

Statistical Analysis

The maximum observed serum levels of TNF- after LPS and susalimod administration were compared with the C_max obtained in the control group using the one-sided t test (Microsoft Excel 7.0).

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Pharmacokinetics of Susalimod. The proposed three-compartment PK model was fitted to the observed serum data of susalimod (Table 1; Fig. 2). Maximum serum levels of susalimod were observed at 15 (first sampling time point) or 30 min after dose administration. The model showed an extensive biliary excretion with a maximum bile clearance of 76 ml/min · kg. Bile clearance appeared to be saturated already at relatively low susalimod concentrations (K_m = 70 µM). Hence, during the first hour after the 25-mg/kg dose ~30% of the dose was estimated to reside in the gut, whereas the corresponding values after the 200-mg/kg dose was only between 6 and 7%. The estimated volume of distribution of the serum and tissue compartments was of the same magnitude, 0.26 and 0.37 l/kg, respectively. In the latter, slowly equilibrating compartment, peak concentrations of susalimod were estimated to be reached between 120 and 180 min.

View larger version (18K): [in this window] [in a new window]   Fig. 2.   Nonlinear regression fit (lines) of the PK model to observed serum concentration data of susalimod following different i.p. doses of susalimod (0, 25, 50, 100, and 200 mg/kg) to female CD-1 mice. A simultaneous fit to all individual data from all doses was performed.

Pharmacodynamics and PK-PD. Experimental data indicated that coadministration of susalimod and LPS leads to a dose-dependent reduction in the TNF- exposure with respect to both C_max and area under the serum concentration-time curve (AUC) (Table 2; Fig. 3). The interindividual variability in cytokine serum levels was very large with coefficient of variations generally in the order of 65% (in the treatment groups as well as the control group). A statistically significant decrease in C_max of TNF- relative to the control group was found at the 100- and 200-mg/kg dose level of susalimod.

View this table: [in this window] [in a new window]   TABLE 2 TNF- serum exposure parameters following concomitant i.p. administration of LPS (mg/kg) and different doses of susalimod (0, 25, 50, 100, and 200 mg/kg) to female CD-1 mice

View larger version (18K): [in this window] [in a new window]   Fig. 3.   Nonlinear regression fit (lines) of the PK-PD model to serum concentration data of TNF- after concomitant i.p. administration of LPS and different doses of susalimod (0, 25, 50, 100, and 200 mg/kg). A simultaneous fit to all individual data from all doses was performed. Only mean observed TNF- levels are shown.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

Administration of LPS to female CD-1 mice induced high serum levels of the proinflammatory cytokines TNF- and IL-6. We found that coadministration of LPS and susalimod lead to a dose-dependent reduction in the TNF- exposure, whereas almost no effect on IL-6 was seen. In addition, 200-mg/kg susalimod induced elevated serum levels of the anti-inflammatory cytokine IL-10. These findings are similar to the effect of the chemical analog sulfasalazine on circulating cytokines in RA patients (Danis et al., 1992). During 6 months of sulfasalazine treatment, a clear decrease in circulating IL-1 and TNF- was demonstrated, whereas no effect on circulating IL-6 was seen. Hence, our results indicate that susalimod has a potential use as a pharmacologically active DMARD.

As previously mentioned, LPS administration in the mouse is known to very rapidly induce TNF- synthesis in monocytes. This process has been described to occur by the following chain of events: 1) LPS binds to LPS-binding protein; 2) the LPS-binding protein complex interacts with CD14 on the cell surface; 3) following signal transduction, the TNF- gene is transcribed to TNF- mRNA; 4) mRNA is either degraded or translated into a precursor protein monomer of TNF-; and 5) precursor TNF- is inserted into the cell membrane, modified by membrane enzymes (trimer formation), and released as mature TNF- (Giroir, 1993; Remick, 1995; Su et al., 1995). The successful fit of the proposed PK-PD model to the experimental data indicated an inhibitory effect of susalimod on the synthesis rate of TNF-, although the present study design did not allow a more specific identification of which step in the LPS-induced TNF- synthesis that was affected. Theoretically, either of the following mechanisms is possible: inhibition of the transcription to TNF- mRNA [as with glucocorticoids (DeForge et al., 1990) and pentoxyfylline (Strieter et al., 1988; DeForge et al., 1990)], inhibition of the translation of mRNA to precursor TNF- [as with glucocorticoids (Han et al., 1990)], or induction of the mRNA degradation [as with thalidomide (Moreira et al., 1993)].

The biological relevance of the proposed PD model was supported by the estimated duration of the TNF- synthesis (~70 min), which is in good agreement with a previous finding, indicating that the expression of TNF- mRNA is very rapid but transient (Wollenberg et al., 1993). TNF- mRNA levels have been reported to peak already at 15 min after LPS challenge and then to decrease to baseline at 1 h. Also, the predicted peak serum levels of TNF- at ~90 min after LPS administration are in accordance with previous in vivo data in the mouse (Remick et al., 1989).

Apart from describing the concentration-effect relationship and, at least to some extent, having contributed to our understanding of the mechanism of action, this PK-PD model also may be used for prediction of outcomes because the study was designed so that dose-independent parameter estimates could be obtained. The doses used allowed an investigation of the PK of susalimod over a wide concentration range and also covered a large part of the pharmacological effect intensity interval. The nonlinear kinetics behavior of susalimod was modeled by a Michaelis-Menten elimination model. It was estimated that the maximum bile clearance was 76 ml/min · kg. Although previously reported clearance values in the mouse, calculated from AUC following i.v. dosing, are much lower (3.6 ml/min · kg at 50 mg/kg and 7.4 ml/min · kg at 25 mg/kg), clearance data obtained in the dog (1.6 ml/min · kg at 50 mg/kg while 15-fold higher, 20 ml/min · kg at 5 mg/kg) (Påhlman et al., 1998) support the conclusions from the present model that clearance may be very high at low susalimod concentrations (K_m). However, the accuracy of the estimates of V_max and K_m may be questioned because they were found to be sensitive to the assumption of equal rates of absorption for the i.p. administered dose and the enterohepatically reabsorbed drug. A more complex model with the rate of reabsorbed drug modeled as a separate parameter was tested, but was rejected because of the goodness-of-fit criteria stated previously. It should be noted, however, that the more complex PK model generated almost identical PK-PD parameter estimates as the model presented herein (data not shown).

The fact that total concentration of susalimod was used in the PK data analysis and not the unbound concentration may have influenced the results because the possible saturation in binding to plasma albumin was not taken into account. As previously mentioned, susalimod is very highly bound in plasma (>99%) in both animals and humans. In vitro protein-binding studies in mouse plasma have shown that the fraction unbound increased from 0.78% at 25 µM to 0.91% at 250 µM, and one can anticipate a further decrease in binding as the susalimod concentration gets close to and exceeds the plasma albumin levels.

Nevertheless, one can conclude that the proposed PK-PD model accurately described the concentration-effect relationship between the immunomodulating agent susalimod and the experimentally induced elevation of proinflammatory cytokine TNF- serum levels. The results suggest that susalimod inhibits the rate of TNF- synthesis with an estimated IC₅₀ of 293 µM in the mouse.