The reversal effect of prothrombin complex concentrate (PCC), activated PCC and recombinant activated factor VII against anticoagulation of Xa inhibitor

This is an in vitro study where the ability of PCC, aPCC and rFVIIa to reverse the effect of rivaroxaban was tested in blood collected from patients treated with rivaroxaban.

Fifty patients treated with therapeutic doses of rivaroxaban for various approved indications and 40 healthy controls, without previous history of vascular disease, were recruited in the study. Patients between 18 and 85 years of age who had taken rivaroxaban for more than two months were eligible. Controls were recruited from the same age group. Ongoing treatment with anti-platelet drug(s) and/or non-steroidal antiinflammatory drug(s) was an exclusion criterium for both patients and controls. All participants gave written informed consent, and the study was approved by the Norwegian regional committee for medical and health research ethics.

We evaluated the following reversal agents in this study: 4-PCC (Cofact®, Sanquin, Amsterdam, the Netherlands), aPCC (FEIBA®, Baxter AG, Vienna, Austria) and rFVIIa (Novoseven®, NovoNordisk, Copenhagen, Denmark). The concentrations included in this study were chosen to imitate 80%, 100% and 125% of the doses suggested for clinical use in case of a major bleeding in a patient treated with a DOAC according to existing guidelines. For PCC the suggested 100% dose is 40 IU/kg, aPCC 50 IU/kg and rFVIIa 90 μg/kg [22]. The drugs were dissolved in sterile water to stock solutions of 34 IU/mL PCC, 34 IU/mL aPCC and 68 μg/mL rFVIIa. Doses of the spiked haemostatic agents were calculated assuming that an adult had 65 mL blood/kg.

Blood was collected from an antecubital vein of the patients through a 21Gx19 mm butterfly needle (Vacuette® Greiner Bio-One GmbH, Kremsmunster, Austria) with minimal use of stasis. The first 2-4 mL of blood was discarded. The blood collection tubes for the measurements of thrombin generation and thromboelastometry (0,109 M citrate Monovette®, Sarstedt, Nümbrecht, Germany) were manually prefilled with Corn Trypsin Inhibitor (CTI) (Haematologic Tecnologies Incorporates, Essex Junction, VT, USA) at a final concentration of 20 μg/mL. Test tubes that were not filled completely were discarded. For anti-FXa activity measurements we used 4.5 mL Vacutainer® tubes (Becton-Dickinson, Franklin Lakes, NJ, USA) containing 0.5 mL 0.109 M buffered citrate without CTI. The blood sampling was performed at the time of presumed peak concentration of rivaroxaban in the patients, about 2 h after the drug intake.

For measurements of rivaroxaban concentration by an anti-FXa activity assay, citrated plasma was obtained after centrifugation for 15 min at 2000 g in RT. The supernatant was carefully collected and stored at -80 °C for 2–3 months before measurements of anti-FXa activity were performed.

Whole blood containing CTI for measurements of thromboelastometry and thrombin generation from each patient was pooled and divided into 10 aliquots of 5 mL. Haemostatic agents were added in the doses mentioned above, and one aliquot was always left untreated to represent the baseline value.

Aliquots of whole blood spiked with three different haemostatic agents at increasing concentrations and the untreated aliquot were incubated at 37 °C for 30 min. Then the samples were further subdivided. Platelet-poor plasma (PPP) was obtained by centrifugation for 13 min at 12000 g in RT, and the supernatant was carefully collected. PPP was immediately frozen and stored at -80 °C for 1–3 months before measurement of thrombin generation. The remaining whole blood was incubated at 37 °C for another 30–90 min before measurements by thromboelastometry were performed.

To measure rivaroxaban concentration in citrated plasma, an anti-FXa activity method calibrated for rivaroxaban was performed on STA-R Evolution® coagulometer (Diagnostica Stago S.A.S., Asnières sur Seine, France) [23] according to the manufacturer’s instructions.

Thrombin generation was measured in PPP using the Calibrated Automated Thrombogram (CAT) (Diagnostica Stago, Asnière, France) with the Thrombinoscope software (Thrombinoscope BV®, Maastricht, The Netherlands) [24, 25]. PPP, supplemented with the three different reversal agents in three different concentrations, were run in triplicates. The thrombin generation parameters lag time, peak of maximum thrombin concentration, velocity index and the total amount of thrombin generated, i.e. endogenous thrombin potential (ETP), were recorded. The PPP reagent containing 5 pM TF and 4 μM phospholipids was used to initiate thrombin generation.

CTI-containing whole blood (with reversal agents in three different concentrations) were run in duplicates and the clotting time (CT; seconds), clot formation time (CFT; seconds), maximum velocity (MaxV; mm/s), area under curve (AUC) and maximum clot firmness (MCF; mm) were measured by ROTEM® (TEM Innovations, Munich, Germany) with low tissue factor activated ROTEM [26]. Prior to measurements, the plastic test cups were prepared with 40 μL buffer (a mixture of equal parts of buffer 1: 20 mM Hepes, 150 mM NaCl, pH 7.4 and buffer 2: 20 mM Hepes, 150 mM NaCl, 200 mM CaCl₂, pH 7.4). Recombinant relipidated TF (Innovin®, Dade Behring, Liederbach, Germany) diluted in a total volume of 20 μL of buffer 1 was also added. To initiate the reaction, whole blood (280 μL) was added and the total volume of reagents and whole blood in each cup was 340 μL. The final TF dilution was 1:70 000, corresponding to a theoretical concentration of 0.35 pM.

The Spearman’s rank correlation coefficient was used when assessing the relationship between rivaroxaban concentration and coagulation parameters. The data are expressed as mean value with a 95% confidence interval (CI 95%) or one standard deviation (SD).

The mean rivaroxaban concentration in the patient group was 216.7 ng/mL (95% CI 188.2–245.3). Rivaroxaban concentrations were compared to thromboelastometry parameters in whole blood and to TGA parameters in PPP. Only data for CT and ETP are shown. The Spearman’s correlation coefficient between rivaroxaban concentration and CT was 0,68 (p < 0.005) (Fig. 1a). There was also a linear negative correlation between rivaroxaban levels and ETP in PPP (r = −0.72; p < 0.005) (Fig. 1b).

Rivaroxaban affected all thrombin generation parameters in PPP. Mean lag time was prolonged more than 3-fold relative to untreated controls (mean difference 6.5 min, 95% CI 5.7–7.3). Peak concentration was reduced by almost 90% (mean difference 167.1 nM, 95% CI 161.6–172.7), Velocity Index (VI) by approximately 97% (mean difference 58.2 nM/min, 95% CI 57.8–58.6) and ETP by approximately 40% (mean difference 532.9 nM*min, 95% CI 416.0–649.8) (Fig. 2).

The haemostatic agents improved the thrombin generation parameters at varying degrees. PCC in a 100% dose did not cause a significant shortening of the lag time (mean difference 1.6 min, 95% CI 1.1–2.2). rFVIIa in 100% dose (90 μg/kg) was more effective than PCC in shortening the lag time (mean difference 6.6 min, 95% CI 5.7–7.4) but this effect was not significantly different from aPCC 100% dose which shortened lag time by 50% (mean reduction 5.5 min, 95% CI 4.7–6.3) (Fig. 3a). PCC in a 100% dose almost doubled peak concentration (mean difference 37.7 nM, 95% CI 29.5–45.9), and so did the 100% dose of rFVIIa (mean difference 36.1 nM, 95% CI 30.2–42.0). However, both those agents were less effective than aPCC 100% which caused an increase in peak concentration by 400% (mean difference 83.9 nM, 95% CI 71.4–96.4) (Fig. 3b). The velocity index (VI) was increased by 240% by PCC 100% (mean difference 3.0 nM/min, 95% CI 2.3–3.8), and by 400% by rFVIIa 100% (mean difference 6.2 nM/min, 95% CI 4.6–7.9). aPCC in the 100% dose had a significantly better reversal effect increasing VI than PCC and rFVIIa and increased VI by 800% (mean difference 12.9 nM/min, 95% CI 9.6–14.6) (Fig. 3c). The total amount of thrombin generated, ETP, was increased by the 100% dose of PCC by 130% (mean difference 802.5 nM*min, 95% CI 649.2–955.8), and by the 100% dose of rFVIIa by 80% (mean difference 452.7 nM*min, 95% CI 365.7–539.7). The 100% dose of aPCC had a more pronounced effect than PCC and rFVIIa and increased ETP by 235% (mean difference 1382.5 nM*min, 95% CI 1203.1–1561.9) (Fig. 3d).

There was not a dose–response relationship for any of the reversal agents on any of the TGA parameters. aPCC 125% of the suggested dose in clinical use reversed the TGA parameters in a similar manner to the 80% dose aPCC (lag time: p = 0.99, peak concentration: p = 1.0, velocity index: p = 0.99, ETP: p = 0.6). Furthermore the reversing effect of the lowest dose aPCC (80%) was more pronounced than the highest dose of PCC (125%) in all TGA parameters (p < 0.005) (Fig. 3a-d).

Compared to controls, rivaroxaban doubled the CT (mean difference 674.4 s, 95% CI 527.8–820.9) (Table 2). PCC shortened the rivaroxaban-induced prolongation of CT by 25% (mean difference 386.9 s, 95% CI 263.0–510.8) and there was not a significant difference between the 80%, 100% and 125% doses (32 IU/kg, 40 IU/kg or 50 IU/kg). Adding rFVIIa caused a shortening of the CT by approximately 40% (mean difference 636.6 s, 95% CI 523.4–747.7) and aPCC shortened CT by 60% (mean difference 892.1 s, 95% CI 762.2–1022.1). There was not a significant difference between rFVIIa and aPCC in a 100% dose nor between the highest doses of these haemostatic agents. However, aPCC was the only haemostatic agent that shortened CT down below the level of the controls, and the 80% dose of aPCC was more effective than the highest dose (125%) of PCC (p < 0.005) (Fig. 4a)

Treatment with rivaroxaban (Table 2) and addition of haemostatic agents (data not shown) influenced CFT in the same manner as CT. However, neither rivaroxaban (Table 2) nor the haemostatic agents (data not shown) affected the thromboelastometry parameters MCF, maxV and AUC.

A subgroup analysis was performed on the patients with the highest rivaroxaban concentrations, defined by a rivaroxaban concentration >300 ng/L (mean 400.6 ng/mL, 95% CI 335.3–466.0) (n = 6). Those patients had the longest CT in WB (mean clotting time 2425.6 s, 95% CI 1774.8–3076.3) and lowest ETP in PPP (mean ETP 531.1 nM/min, 95% CI-187.7–1249.9). We found the same pattern as in the main analysis, i.e. that aPCC increased ETP more than PCC (p < 0.001) and rFVIIa (p < 0.001). All three haemostatic agents reduced the CT in the subgroup analysis, but there was not a significant difference between the three drugs (p = 1.0) (Fig. 4b).