Five clinically healthy horses were used in this trial. They were group-housed at the Equine Park at Cornell University. The horses were transported the day prior to drug administration to the Cornell University Hospital for Animals facilities and placed in individual stalls. They consisted of one gelding and four mares, ranging from 15 to 19 years of age, 511–626 kg bodyweight, and were from four different breeds (two Oldenburgs, and one each of Holsteiner, Quarter Horse, and Warmblood).

For the preparation of the oral formulation, apixaban tablets (Eliquis®, Bristol-Myers Squibb, Pfizer, New York, NY, USA) were crushed with a mortar and pestle and the powder was mixed with 70 g of molasses (Herdlife™, New Roads, LA, USA). The paste was transferred to a 60 mL syringe (Becton-Dickinson) and stored at room temperature until administration. For IV formulation, crushed tablets were dissolved in a vehicle consisting of 10% v/v N,N-dimethylacetamide (EMD Millipore, Billerica, MA, USA), 17% v/v propylene glycol (Sigma-Aldrich), 17% v/v dimethyl sulfoxide (DMSO, Sigma-Aldrich), and 56% v/v deionized water (MilliQ® water, Q-Pod®, EMD Millipore). After vigorous vortexing, the formulation was sterile filtered through a 0.22 μM cellulose acetate filter (Corning Inc., Corning, NY, USA) then drawn into 60 mL syringes. The IV formulation was prepared 24 h before administration and stored at room temperature. We aimed for a final concentration of 2.5 mg/mL apixaban in the IV formulation. Since each tablet of Eliquis® weighed 210 mg, equal weights of lactose (Blackburn Distributions, Nelson, Lancashire, UK) and microcrystalline cellulose 102 (Blackburn Distributions) were crushed to create a placebo, which was similarly mixed with molasses or vehicle diluents for oral and IV administration, respectively, and stored as for the apixaban formulations. To ensure that the vehicle was not toxic, safety testing was performed on a trial IV dose given to a different horse. The horse had no adverse clinical reactions and there were no changes in clinical pathologic test results 24 h after IV administration of the bolus of dissolved apixaban.

In this double-blinded placebo-controlled cross-over study, apixaban or placebo was given either per os (PO) or by IV injection to horses at 0.2 mg/kg, followed by a 14-day washout between treatments. This dose was chosen based on the pharmacokinetic parameters after oral and IV administration to cats, which resulted in a C_max of 74 and 460 ng/mL, respectively (25). The first and last treatments for each horse were IV administration of drug or placebo (Figure 1). Personnel, who were not involved in preparing and administering the drugs, data acquisition, or data analysis, randomized the oral or IV treatments for each horse between placebo and apixaban. Investigators were blinded to each treatment until the conclusion of the study.

Blood samples were collected before drug administration (time 0) and at 3 (IV) or 15 (PO) min, then at 30, 45 min, 1, 2, 3, 4, 6, 8, and 24 h after drug administration into 6-mL syringes (Monoject Syringe, Covidien Ltd, Mansfield, MA, USA) prefilled with 3.8% citrate anticoagulant (Sigma-Aldrich, St Louis, MO, USA) maintaining a ratio of 9:1 blood:citrate. The citrate-anticoagulated blood was used to prepare platelet-poor plasma (PPP) for measurement of UPLC-MS concentrations, anti-Xa activities and dilute prothrombin times (dPT) and platelet-rich plasma (PRP) for flow cytometric analysis of ex vivo platelet activation in response to exposure to EHV-1. Blood was also collected into EDTA- and non-anticoagulant samples for clinical pathologic testing before and 24 h after each treatment (Figure 1). With the exception of 24 h samples, blood was collected at each time point through an indwelling jugular catheter (14 G × 15 cm, Milacath®, MILA International, Florence, KY, USA) after removing 20–30 mL of blood. Patency of the catheters was maintained by flushes with sterile saline solution (0.9% sodium chloride injection USP, Baxter Healthcare Corporation, Deerfield, IL, USA) after each blood draw. The catheter was removed after collection of the 8 h sample. For IV administration, a second catheter was placed in the opposite jugular vein for bolus drug delivery and removed immediately after drug administration and flushing of the catheter with sterile saline. For the 24 h sample, blood was collected from the jugular vein (the vein contralateral to that used for the indwelling catheter) into a 3.8% citrate anticoagulant prefilled 6-mL syringe via an 18 G needle (Monoject Veterinary Needle, Covidien Ltd, Mansfield, MA, USA) and into EDTA- and non-anticoagulant vacutainers (Becton-Dickinson, Franklin Lakes, NJ, USA) via a vacutainer needle (21 G, Becton-Dickinson Eclipse™ Vacutainer® with pre-attached holder).

To measure the extent to which apixaban was bound to protein, previously frozen PPP from each horse from the first baseline sample was thawed in a 37°C waterbath and centrifuged at 16,000 × g for 5 min at room temperature. The supernatant was filtered with a 0.22 μM syringe filter (Corning Inc.). The PPP was then spiked with apixaban at 200 and 1,000 nM (equivalent to 92 and 460 ng/mL) using the same IV preparation that was administered to the horses. Half of the sample was subsequently frozen at −20°C and the rest was ultracentrifuged at 16,000 × g for 30 min in a molecular exclusion device (Amicon Ultra 30k filter, Millipore), to remove the major plasma proteins. The ultrafiltrate was then frozen at −20°C. Both the spiked plasma and the ultrafiltrate of the spiked plasma were analyzed for apixaban using UPLC-MS. The percentage of protein binding was calculated by the difference between the apixaban concentration in the spiked plasma and the ultrafiltrate divided by the concentration in the spiked plasma.

This was performed as previously described (33). Analysis by UPLC-MS/MS was performed on a 2D UPLC system (Waters Acquity UPLC H-class with 2D Technology System, Waters GmbH, Eschborn, Germany) directly coupled to a Xevo TQ-S tandem mass spectrometer (Waters GmbH, Eschborn, Germany) which was operated in electrospray positive ionization mode. The system control and data acquisition were performed using MassLynx NT 4.1 software with automated data processing by the MassLynx QuanLynx program provided with the instrument. The lower limit of detection for all DOACs of the UPLC-MS/MS method was < 0.2 ng/mL. Since DOAC values < 7 ng/ml were not clinically relevant, all time points below this value were considered zero.

The anti-Xa activity of apixaban was measured in citrate-anticoagulated PPP by the Comparative Coagulation Laboratory in the Animal Health Diagnostic Center at Cornell University. The PPP was harvested from the supernatant of PRP after high-speed centrifugation (13,000 × g, Accuspin Micro, Thermo Scientific, Rockford, IL, USA) for 5 min. The PPP was frozen at −20°C and assays were performed in batches. The PPP was thawed at 37°C in a water bath before analysis with an automated coagulation analyzer (STA® Compact, Diagnostica Stago, Parsipanny, NJ, USA). The assay is configured with a bovine activated Factor X reagent added in excess to the test plasma and a chromogenic substrate of Factor Xa (STA®-Liquid anti-Xa, Diagnostica Stago). In this assay, residual, uninhibited Factor Xa cleaves the chromogenic substrate such that the inverse of the color change in the reaction mixture is proportional to the drug concentration in the test plasma. Results are expressed as ng/mL anti-Xa activity, based on the calibration standard containing known apixaban concentrations in human plasma (STA®-Apixaban Calibrator, Diagnostica Stago). Assay controls, consisting of apixaban spiked-human plasma (STA®-Apixaban Controls) were measured before each batch of test samples.

In an attempt to determine pharmacodynamic profiles of apixaban after IV administration, a dPT assay was performed on PPP samples at all time points after IV injection of apixaban or placebo. The assay was configured with a rabbit brain thromboplastin reagent (Thromboplastin LI, Helena Diagnostics, Beaumont, TX) prediluted 1:2 in an imidazole buffer containing 0.025 M calcium chloride. The reagent was diluted to enhance its sensitivity to detect the presence of apixaban's anticoagulant effect (34, 35). Test results were reported as clotting time in seconds.

Pharmacokinetic analysis was performed using a two-compartment open model with weighting of 1/y² (Phoenix®, WinNonlin® software, Certara Inc., St. Louis, Missouri). Compartmental analysis of the data after administration of apixaban was calculated according the following formula:

where C is the apixaban concentration, A is the distribution phase y-axis intercept, e is the base of the natural logarithm, t is time after administration, α is the distribution rate constant, B is the elimination phase y-axis intercept, and β is the elimination phase rate constant (terminal phase). Secondary parameters calculated include distribution (α) and elimination (β) half-lives (T½), microdistribution rate constants, area-under-the-curve (AUC), apparent volume of distribution at steady-state (V_SS), systemic clearance (CL), and mean residence time (MRT).

The pharmacodynamic analysis was performed by plotting the drug concentration (apixaban measured by UPLC-MS) on the x-axis and response (dPT) on the y-axis to determine if a relationship existed.

We used flow cytometric measurement of agonist-induced P-selectin expression as an ex vivo test for the inhibitory activity of apixaban on thrombin generation. This was only done on samples collected before and 2 and 24 h after drug administration (Figure 1). We chose the 2 h time point for flow cytometric analysis, because this was the time of peak concentration achieved after oral administration and concentrations were still high at this time, albeit not at peak, after IV administration in cats (25). Flow cytometric analysis was done as we have previously described (36). In brief, the citrate-anticoagulated blood was processed within 30 min of collection. Red blood cells were allowed to settle by gravity sedimentation at room temperature for 20 min. Then PRP was prepared by centrifugation of the resulting leukocyte-platelet-rich plasma at 250 × g at 21°C for 10 min. Thrombin at 0.15 U/mL (Sigma-Aldrich), bovine factor Xa at 0.1 and 1 μg/mL (Haematologic Technologies, Essex Junction, VT, USA) and two strains of EHV-1 were used as platelet agonists, and phosphate-buffered saline (PBS, Mediatec, Manassas, VA, USA) was used as a negative control. The strains of EHV-1 were Ab4 (propagated in rabbit kidney 13 cells) and RacL11 (propagated in equine kidney cells) and they were used at low and high concentrations equivalent to 1 × 10⁶ plaque forming units (PFU)/mL and 5 × 10⁶ PFU/mL, respectively. Both strains were isolated from cell lysates on a sucrose cushion via ultracentrifugation as we have previously described (12). One harvested isolate of each strain was used for the entire study.

Platelet-rich plasma was diluted at a 1:5 ratio in binding buffer (10 mM HEPES, 140 mM NaCl, 2.5 mM calcium chloride, 1 mM glycine–proline–arginine–proline, pH 7.4, all chemicals from Sigma-Aldrich), then was incubated for 10 min with the various agonists. The platelets were then labeled with an Alexa647-conjugated antibody against P-selectin (final concentration of 66 ng/mL, clone Psel.KO.2.7, Novus Biologicals LLC, Littleton, CO, USA) for 10 min at room temperature in the dark. The reaction was quenched with binding buffer and the samples were immediately analyzed with a flow cytometer (BD FACSCalibur, BD Biosciences, San Jose, CA, USA). For analysis, platelets were gated on their forward and side scatter characteristics and the percentage of P-selectin-positive platelets determined from histogram plots.

This was performed in the Clinical Pathology laboratory in the Animal Health Diagnostic Center at Cornell University and was done to determine if there were any apixaban-related effects on hematologic or select biochemical test results after PO or IV administration. Automated hemogram results were obtained from the EDTA-anticoagulated blood with a hematology analyzer (ADVIA 2120, Siemens Healthcare Diagnostics Inc., Tarrytown, NJ, USA). To verify the automated platelet counts, modified Wright's-stained blood smears were evaluated for platelet clumps by trained medical technologists. Biochemical analysis for renal analytes (serum urea nitrogen and creatinine concentrations), liver enzymes (sorbitol dehydrogenase, glutamate dehydrogenase, aspartate aminotransferase and gamma glutamyl transferase activities), protein-related analytes (total protein, albumin, and globulin concentrations), and the acute phase protein, serum amyloid A, were performed with an automated chemistry analyzer (Hitachi P modular, Roche Diagnostics, Indianapolis, IN, USA) using manufacturer's reagents, with the exception of sorbitol dehydrogenase (Sigma-Aldrich), glutamate dehydrogenase (Randox Laboratories Ltd, Antrim, UK), and serum amyloid A (LZ test Eiken-SAA, Mast House, Merseyside, UK).

Data was tested for normality using a Shapiro-Wilk normality test. Comparison of the mean or median of 2 groups was conducted with a paired t-test or Wilcoxon matched-pairs signed rank test, respectively. Multiple means or medians were compared with one-way ANOVA corrected for multiple comparisons with Tukey's post-hoc method or a Friedman test with Dunn's post-hoc method, respectively. Pearson correlation analysis was used to assess the association between apixaban plasma concentrations determined by UPLC-MS and the anti-Xa activity assay or the dPT. Alpha was set at 0.05. Statistical software used was Prism 7.03 for Windows (GraphPad Software Inc, La Jolla, CA, USA).

Liquid chromatography data revealed that apixaban was not detected in plasma after oral administration. The mean apixaban concentration of the 5 IV formulations was 1.8 mg/mL (range, 0.84–2.6 mg/mL) reflecting a 28% loss of drug compared to the intended concentration of 2.5 mg/mL. Therefore, the average administered dose given to the 5 horses was 0.15 mg/kg instead of 0.2 mg/kg. A two-compartment model was found to be the best fit after IV administration for all five animals (Figure 2). The mean ± standard deviation (SD) plasma concentration of apixaban peaked at 733 ± 136 ng/mL at 3 min and decreased to undetectable levels by 8 h in all horses. The drug had a short elimination half-life of 1.3 ± 0.2 h and a low volume of distribution (VD_SS 93.3 ± 75.6 mL/kg) (Table 1). The mean percentage of protein binding for apixaban was 92.4 ± 0.5 and 98.5 ± 0% for 92 and 460 ng/mL, respectively.

There was no discernible relationship between the dPT and plasma apixaban concentrations measured by the UPLC-MS (Figure 3), precluding the use of this coagulation assay as the response for measuring pharmacodynamic parameters of apixaban. Individual horses showed substantial variability in their dPT over time, which contributed to the lack of an observed relationship.

There was no decrease in the median percentage of platelets expressing P-selectin in PRP when activation was induced by a high concentration of bovine factor Xa or low or high concentrations of the RacL11 EHV-1 strain at 2 h after PO or IV administration of apixaban (Tables 2, 3; Figures 4, 5). A decrease in the median percentage of platelets expressing P-selectin was seen in PRP exposed to low concentrations of bovine factor Xa and both concentrations of the Ab4 EHV-1 strain at 2 h after IV administration of apixaban. However, similar changes were seen in PRP stimulated with high concentrations of Ab4 at 2 h in animals given the IV placebo (Figure 5). In addition, platelets were either not reactive or minimally reactive to the low concentration of bovine factor Xa or both concentrations of Ab4 at 24 h, regardless of whether placebo or apixaban were administered PO or IV (Tables 2, 3; Figures 4, 5). These findings argue against any drug effect on platelet activation at the 2 h time point. Since median concentrations may mask individual results, we also examined changes in the individual horses with the low concentrations of each agonist after IV administration only. This data showed that platelet reactivity was reduced at 2 h in the PRP of four apixaban-dosed horses with the three agonists at low dose, with one horse (horses 5) showing decreases with all three agonists. In contrast, no decreases in platelet reactivity were seen in placebo-treated horses. However, in two out of the four horses, the changes in P-selectin expression were mild or still present at 24 h, making it difficult to attribute the decreases at 2 h to a drug effect (Figure 6). At the 2 h time point, the apixaban concentration ranged from 44 to 74 ng/mL in individual horses after IV administration. When we spiked equine PRP with similar concentrations of apixaban ex vivo, we did not see any inhibition of platelet activation (data not shown), further support for a lack of drug-induced inhibition of platelet reactivity at the 2 h time point.

The apixaban concentrations measured with the anti-Xa activity assay showed excellent correlation to the UPLC-MS method (r² = 0.9997, P = < 0.0001, Figure 7). No change in anti-Xa activity was seen with the oral or IV placebo or oral dosing of apixaban. The median anti-Xa activity differed significantly from baseline only for the first hour after IV administration of apixaban (Table 4).

There were no clinically apparent adverse effects associated with the administration of apixaban with oral and IV formulations in any horse. There were no significant changes in any hematologic test result 24 h after PO or IV administration of placebo or apixaban (Table 5). Some individual horses had test results that were slightly above or below the established reference intervals, but this likely represented normal biological variability. A low platelet count was seen in 1 horse at two different timepoints, but platelet clumps were noted in the blood smears. Similarly, there were no significant changes in any of the chosen biochemical test results 24 h after PO or IV administration of placebo or apixaban (Table 6). Individual horses showed variability in biochemical test results, with a few horses having transiently increased liver enzyme activities or decreased albumin concentrations at one or both timepoints. Again, we attributed this to normal biological variation.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.