Correlating prothrombin time with plasma rivaroxaban level

Recently published guidelines from the British Committee for Standards in Haematology suggest that the prothrombin time (PT) can be used for urgent determination of anticoagulation intensity with rivaroxaban if a reagent with a known sensitivity is used (Baglin et al, 2012). However, it has been established in experiments using normal plasma spiked with varying concentrations of rivaroxaban that different PT reagents may have very different sensitivities to rivaroxaban (Samama et al, 2010; Douxfils et al, 2012). The clinical implications of this variation in PT sensitivity to rivaroxaban is evident from the recent report by Van Veen et al (2013), who described normal PTs in the presence of therapeutic rivaroxaban levels in a patient with renal impairment. Comparing the PT using Innovin® (Siemens, Marburg, Germany) and Thromborel S® (Siemens) in a patient commenced on rivaroxaban 15 mg twice daily for deep vein thrombosis against the rivaroxaban plasma concentration measured by the Biophen DiXal anti-Xa assay, they demonstrated normal PT results despite therapeutic plasma rivaroxaban levels. They concluded that PT results should be interpreted with caution when assessing coagulation intensity for patients on rivaroxaban (Van Veen et al, 2013). If available, a specific anti-Xa assay should be used for patients presenting with major bleeding or when requiring emergency surgery. Mueck et al (2011) have also demonstrated variation using Neoplastine® (Diagnostica Stago, Asnieres-sur-Seine, France) to measure the PT against rivaroxaban concentration. However, many coagulation laboratories will not have 24-h access to a rivaroxaban-calibrated anti-Xa assay and will inevitably rely on their routine coagulation screen PT and activated partial thromboplastin time (APTT) results. Therefore it is essential that laboratories have knowledge of their reagents sensitivity to rivaroxaban. We studied the correlation between the rivaroxaban concentration (measured by Liquid Anti-Xa chromogenic assay, (Instrumentation Laboratory Company, Bedford, MA, USA) run on an IL TOP 700) and PT and APTT. The anti-Xa assay was calibrated using lyophilized standard rivaroxaban plasmas (Hyphen Biomed, Neuville-sur-Oise, France) to create a rivaroxaban assay linear between 0 and 400 ng/ml. Peak rivaroxaban plasma concentrations are considered to be in the range of 100–400 ng/ml, and trough concentrations in the range of 20–150 ng/ml (Baglin et al, 2012). PT (Recombiplastin 2G®, Instrumentation Laboratory Company, laboratory reference range 9–13 s) and APTT (Synthasil®, Instrumentation Laboratory Company, laboratory reference range 27–38 s) were determined on the IL TOP 700. Blood samples (n = 33) were collected at random time points from 31 patients receiving rivaroxaban treatment for a minimum of 2 weeks. Fourteen results were obtained during 15 mg twice daily dosing and 19 during 20 mg once daily dosing. 19 blood samples were taken around the peak plasma concentration (1·0–5·5 h), 13 during trough periods (12–30 h) and for one patient, the sampling time was uncertain. The anti-Xa rivaroxaban concentrations during peak hours were marginally higher than was expected, with a median of 280 ng/ml (range 168–458). The trough levels were as expected, with a median of 57 ng/ml (range 11–215). There were two patients suspected of non-compliance as their anti-Xa levels were inappropriately low for their stated 'peak' level. PT results (range 10–25 s) showed a strong linear correlation with anti-Xa rivaroxaban concentration (correlation co-efficient 0·926) as seen in Fig 1A, while APTT results (range 30–52 s) showed a weaker correlation (correlation co-efficient 0·637, Fig 1B). All the patients (n = 10) who had a normal PT (≤13 s) had rivaroxaban concentrations <90 ng/ml. Six patients with a normal APTT (≤38 s) had an anti-Xa rivaroxaban concentration >100 ng/ml (range 130–441 ng/ml). Our results have demonstrated, in contrast to Van Veen et al (2013), that a normal PT may exclude therapeutic anticoagulation with rivaroxaban, when using Recombiplastin 2G® on an IL TOP 700. This apparent greater sensitivity to rivaroxaban of Recombiplastin 2G®, compared to Innovin®, is consistent with in vitro spiking studies (Samama et al, 2010; Douxfils et al, 2012). To our knowledge this is the first report confirming a similar pattern in plasmas from rivaroxaban-treated patients. There appears to be a stronger correlation with PT, rather than with the APTT, at providing an approximate estimate of rivaroxaban anticoagulant activity. Although this is a relatively small cohort, the findings support the suggestion that a normal PT measured by Recombiplastin 2G® (a reagent used by more than 25% UK National External Quality Assessment Service for Blood Coagulation participants) could infer suitability for surgery or other invasive procedures. These results may only be applicable to Recombiplastin 2G® and caution should be taken in applying these to other reagents. Furthermore, if time allows and a suitable assay is available, a rivaroxaban-calibrated anti-Xa assay may be the most reliable method of determining a low level of rivaroxaban anticoagulant activity prior to invasive procedures, particularly until we have clinical evidence that rivaroxaban-treated patients with a normal PT do not bleed excessively during surgery. The authors have no competing interests. RR wrote the manuscript and RCT and CNB commented on the manuscript. RR, RCT and CNB analysed and interpreted the data. RCT, CNB, CL, GH and MM contributed in designing the research. RCT, CNB and MM provided the samples for research.