Pharmacokinetic/pharmacodynamic modeling for the determination of a cimicoxib dosing regimen in the dog

Non-Steroidal Anti-Inflammatory Drugs (NSAIDs) constitute an essential therapeutic class both in human and veterinary medicines. It is now well established that most side effects of NSAIDs are related to their anti-cyclooxygenase-1 (anti-COX-1) activity while the therapeutic effects are mainly associated with COX-2 inhibition. This has led to the development of preferential or selective COX-2 inhibitors among which are the diarylimidazol derivatives (COXIB class) in an attempt to overcome the side effects especially those of the gastro-intestinal tract (g.i.t). One of these substances was recently developed for therapeutic use in dogs: cimicoxib which is a sulfonamide, like celecoxib, mavacoxib and firocoxib. High in vitro potency of cimicoxib was measured for COX-2 inhibition using a human whole blood assay (half maximal Inhibitory Concentration (IC50) 66nM or 25 ng/mL). In vivo effects were demonstrated in rats in adjuvant-induced arthritis, and in air pouch and hyperalgesia models [1]. In fact, the COXIB class is not homogenous regarding the COX-1/COX-2 selectivity and some of its members such as deracoxib and mavacoxib are classified as preferential and moderately selective COX-2 inhibitors while others like firocoxib and robenacoxib are classified as highly selective COX-2 inhibitors with a more than 100-fold greater potency for COX-2 inhibition and with little or no in vivo inhibition of COX-1 [2]. Usually, with highly selective COXIB compounds, no gastro intestinal tract (g.i.t) ulceration or antiplatelet effects are expected even at a maximum plasma concentration while significant COX-1 inhibition may occur with preferential COXIBs [2]. For example, for deracoxib, the COX-1/COX-2 selectivity index (using a whole blood assay and expressing results as the ratio of IC₅₀) was about 50–60 [3] and gastric ulcerations were reported following some overdosing of deracoxib or as a result of its association with other substances such as other NSAIDs or corticosteroids [4, 5]. This means that a COX-1/COX-2 selectivity index of 50 is not enough to guarantee a margin of g.i.t safety and more highly selective COXIBs are desirable. This is the case of firocoxib with a selectivity index of 384 in a whole blood assay [6] and for which a favorable tolerability was shown in a large scale survey with a withdrawal rate associated with g.i.t side effects as low as 2.9% of dogs (mainly vomiting) and no serious drug-related adverse events [7]. However it was shown that firocoxib was able to slow down wound healing in a canine gastric mucosal injury model but by a mechanism independent of prostaglandin synthesis [8] meaning that a high COX-1/COX2 selectivity is not the only pharmacodynamic (PD) endpoint to consider when discussing NSAID g.i.t tolerability. Another factor of COXIB tolerability is the selected dosage regimen; a dosage regimen depends on two pharmacokinetic (PK) parameters (namely plasma clearance and bioavailability for the extravascular route of administration) and of one PD parameter (namely the efficacious plasma concentration that reflects the drug potency). In addition, the dosage interval should also be rationally determined according to the plasma terminal half-life (which itself depends on plasma clearance and volume of distribution [9]). For the COXIB class, the plasma clearances are very different among substances from a very low clearance for mavacoxib (2.7 mL/kg/h) [10] to a rather high clearance for robenacoxib (810 mL/kg/h) [11] and firocoxib (462 ml/kg/h) [6] which explains the large differences in half-lives namely 17.3 days (beagle dogs) for mavacoxib, 0.63 h (beagle dogs) for robenacoxib and 5.9 h (mixed-breed dogs) for firocoxib. Such differences are reflected into the dosage regimen (dose and dosing interval), with mavacoxib marketed at 2 mg/kg at one month intervals vs. 5 mg/kg daily for firocoxib and 1-2 mg/kg daily for robenacoxib. It is expected that a very short half-life of about 1 h cannot maintain steady plasma concentrations over the entire daily dosage interval while a very long terminal half-life is ineluctably associated with a minimal delay (about 3–4 times the duration of half-life) to achieve steady state conditions. In addition a population PK survey for mavacoxib in osteoarthritic dogs, showed wide between-subject variability with a typical terminal half-life (population mean) of 44 days, but for some dogs (5% of the population) it exceeded 80 days with extreme values up to 140 days. It was shown for mavacoxib that the PK differed considerably between young adult beagle dogs and the typical geriatric large-breed osteoarthritic patient [12] indicating that considering only preclinical PK in young beagle dogs can be very misleading when rationally determining a dosage regimen for a COXIB. In the present paper this point was specifically addressed for cimicoxib by investigating the PK profiles in breeds other than beagle dogs.

The present study reports how the dosage regimen of cimicoxib (Cimalgex ND) was established using PK/PD concepts [13–22] and provides preliminary information on the interbreed variability of cimicoxib PK.

Cimicoxib is a new coxib marketed in veterinary medicine and with a terminal half-life between 2 and 4 h; its pharmacokinetic profile is intermediary between that of robenacoxib (half-life of 1 h) and that of firocoxib (half-life of 7 h). During this experiment in beagle dogs we identified two subgroups of subjects: one group of dogs eliminating cimicoxib from plasma at a rapid rate and a second group of dogs having a three-fold lower plasma clearance ; they were termed by others who observed the same phenomenon for celecoxib [23] extensive metabolizers (EM) and poor metabolizers (PM). The origin and mechanism of these different elimination patterns of cimicoxib were not investigated but are likely to be due to polymorphism of the enzymes cytochrome P450 2D15 (CYP2D15) in dogs, which have been shown to be involved in the polymorphism of canine celecoxib metabolism [23]. For mavacoxib, broad variability in pharmacokinetics was also observed within laboratory beagle dogs with a typical half-life of 15–17 days while some of these beagle dogs had a half-life >30 days. For mavacoxib it was postulated that there was polymorphism of a transporter involved in the biliary clearance even if different sub-populations could not be detected [12]. No equivalent information has been published for firocoxib. In the present experiment, in order to select the most relevant subpopulation of beagle dogs (EM or PM) regarding the future target dog population we conducted a pilot study in field conditions on four different dog breeds of different body size (Anglo-French Hounds, Pointers, Cavalier King Charles Spaniels and Bernese Mountain Dogs) and a rather short terminal half-life (from 2-4 h) was observed for the 40 dogs investigated and we decided that our subpopulation of EM beagle dogs should be considered for the PK/PD dosage regimen determination.

For PK/PD investigations, a reversible kaolin paw inflammation model i.e. an ethically acceptable inflammation model was used to determine the main PD parameters of cimicoxib for a set of clinically relevant endpoints able to reflect the antipyretic, analgesic and anti-inflammatory potential. This model has been validated in our hands and is fully described elsewhere [15]. For the current experiment, the duration of the cimicoxib effect was short enough to avoid having to model the time development of inflammation as was required for meloxicam using the same protocol [15]. In other words, it was assumed that the natural progression of the inflammation following a kaolin administration (increasing then decreasing after a more-or-less well-defined plateau phase) was not a relevant confounding factor to assess the cimicoxib effect. In the model for the placebo period, body temperature, creeping time, and paw withdrawal time still were altered in first 56 h after kaolin administration (spontaneous improvement was 50%, a 100% improvement meaning return to the prekaolin administration condition). This model (inflammation >24 h) used to test daily administration will be preferred to model with short inflammation like sodium urate crystal (inflammation <18 h); model used for robenacoxib and firocoxib [24, 25]. Using a classical PK/PD modelling approach, the cimicoxib efficacy (expressed as a percentage of return to control values) and the cimicoxib potency (measured by the plasma EC₅₀/IC₅₀) were determined. For the subgroup of EM, EC₅₀/IC₅₀ for rectal temperature, creeping speed under a tunnel, vertical ground reaction force and a clinical lameness, scores were between 0.422 and 0.745 nmol/ml i.e. slightly lower than those obtained for meloxicam that were between 0.553 and 1.439 nmol/mL using the same canine model and the same investigated endpoints indicating that cimicoxib is slightly more potent than meloxicam, a reference NSAID in dogs. The efficacy parameters i.e. Imax were similar for cimicoxib and meloxicam (from 70 to 100% for cimicoxib and from 80 to 100% for meloxicam) indicating that both drugs can totally or near totally suppress symptoms associated with the kaolin inflammation (hyperthermia, lameness).

For the analgesic effect, it was shown that cimicoxib has two types of effect: a so-called anti-hyperalgesic effect (i.e. an increase in the pain tolerance threshold which was abnormally lowered by the inflammation) and also an increase to the pain tolerance threshold beyond that recorded during the control (placebo) period suggesting a mechanism of cimicoxib action independent from its anti-inflammatory effect. A similar observation has already been seen with other COXIBs. For example, using a model of peripherally induced inflammatory pain (intraplantar injection of carrageenan) in rats, celecoxib given systemically induced a state of hypoalgesia where the (ipsilateral) nociceptive threshold was raised above basal values. This effect was not observed after treatment with non-selective inhibitors of COX (indomethacin, piroxicam). As naltrexone was able to reverse this hypoalgesia, it was concluded that celecoxib could act centrally via an effect mediated by endogenous opioids rather than by inhibition of PG biosynthesis [26]. Recently in rat, it was demonstrated that analgesic central effects of celecoxib were prevented by selective μ-(β-funaltrexamine) and δ-(naltrindole), but not ĸ-(nor-binaltorphimine) opioid antagonists, after intracerebroventricular administration [27]. This genuine effect of COXIBs should deserve attention because cimicoxib could have some advantages over non-COXIB NSAIDs to treat some (chronic) conditions in which pain is unrelated to inflammation.

The ultimate goal of the present experiment was to assist in the selection of a dosage regimen for subsequent assessment in clinical trials. For this purpose, dose-effect relationships were simulated using PK and PD parameters obtained in the EM beagle dogs because their PK parameters were closer to those of the four non-beagle breed dogs. As explained in the result section, our simulations assumed a linear disposition of cimicoxib. However, the disposition of cimicoxib after an oral administration was subsequently shown to be not linear due to solubility saturation and the results of our simulations should only be considered as indicative. Visual examination of the simulated dose-responses (from 0.1 to 8 mg/kg) with the various PD endpoints (rectal temperature, creeping speed to cross a tunnel, vertical ground reaction force of paw and clinical lameness score), showed that a near-maximal effect was obtained with a dose of 2 mg/kg for all effects but the vertical force and that, above this dose, a further increase was only able to increase the duration of the cimicoxib effect. Visual inspection of the PK/PD simulations clearly showed that a full pharmacological effect of cimicoxib cannot be maintained over a 24 h dosage interval with a single daily dose. However, it was considered that to split the daily dose was not in order given that for other short acting COXIBs such as robenacoxib, a once daily administration (1 mg/kg) was shown to be clinically appropriate; Clinical trials with cimicoxib supported this interpretation (see later) and this is likely due to the fact that our experimental model is more severe that most of the clinical conditions encountered and also to the likely longer persistence of the cimicoxib in the synovial fluid of the target dog population as already shown for robenacoxib [28]. Higher concentration of coxib and longer residence times in inflammatory exudate compare to blood is likely a property of class [1, 29–31] and it is reasonable to assume that it is also true for cimicoxib. To support this single cimicoxib daily dose, the company carried out field trials on peri-operative pain associated with orthopaedic or soft tissue surgery using carprofen as a comparator (4 mg/kg per day). Carprofen is as well-established NSAID with a half-life of about 9 h in dogs [32], i.e. a half-life well-suited for a single daily dose regimen. The non-inferiority of cimicoxib was concluded in comparison to carprofen on pain during the first 24 hours using a composite pain clinical score and a 20% difference as the inferiority limit [33]. For the relief of pain and inflammation associated with osteoarthritis, the non-inferiority of cimicoxib against firocoxib was also demonstrated with regard to the primary efficacy endpoint and an inferiority limit set to 20% [34]. As cimicoxib administered once a day at a dose of 2 mg/kg was shown not to be inferior to carprofen and firocoxib, two drugs having a longer half-life than cimicoxib, the European Agency concluded that although PK/PD data indicated that the duration of the effect was shorter than the proposed dosing interval, this was regarded as acceptable, given that appropriate information to this effect was included in the product literature.

Cimicoxib is an efficacious anti-inflammatory and analgesic drug and thanks to a PK/PD approach, a rational dosage regimen of 2 mg/kg was selected for confirmatory clinical trials allowing in fine the claim that cimicoxib at a daily dose of 2 mg/kg is not inferior to two well-established comparators, namely carprofen and firocoxib.