In this study, rivaroxaban levels were measured by two chromogenic anti-Xa assays and UHPLC-MS/MS method in 243 samples from 182 rivaroxaban-treated Chinese patients with NVAF. Results showed that a wider range of rivaroxaban concentration from Chinese patients in real-world clinical setting compared to the expected range from clinical trial, no matter what kind of method was used. Moreover, chromogenic anti-Xa assays showed a good correlation with the UHPLC-MS/MS over the range of 50–200 ng/mL, but with a trend of underestimation over the whole range.
A large inter-individual variability of rivaroxaban plasma concentration was observed, and the variability was greatest by UHPLC-MS /MS, followed by Biophen DiXal and Zhenyuan. This range is consistent with previous reports in patients or healthy volunteers [15, 18, 19, 27]. Our studies showed a larger inter-individual variability in both peak levels (115.8–600 ng/mL) and trough levels (1.06–261 ng/mL) measured by UHPLC-MS/MS, compared to previously expected ranges for Cmax (178–343 ng/mL) and Ctrough (12–137 ng/mL) for NVAF patients taking 20 mg qd with CrCl ≥ 50 mL/min [2]. The variability of trough and peak concentrations measured in our study was greater, which could be explained by the fact that some patients took a relatively low dose (5 mg qd) and the higher inter-individual variability in patients from real-world clinical practice than in the clinical trial [14].
We observed a high proportion of patients who were not prescribed an inappropriate dose, especially for those with CrCl 50–79 mL/min. Meanwhile, these patients with CrCl 50–79 mL/min had a lower probability of trough concentrations exceeding the expected range, compared to patients with CrCl 15–49 or > 80 mL/min. Lower prescribed NOAC doses in Asian population have been widely reported previously, perhaps due to the consideration of lower body weight, higher proportion of elderly, higher risk of bleeding of Asians[28–31]. Although there was a gap between the rivaroxaban exposure and clinical outcomes till now, some studies showed that a lower rivaroxaban dose did not reduce drug effectiveness and improve the safety in some patients [29, 31, 32].
Chromogenic anti-Xa assays showed a good linear correlation with UHPLC-MS/MS over the quantitative range. However, results showed that there is a systematic underestimation by the chromogenic anti-Xa assays, and the median bias in Biophen DiXal (15.1 ng/mL) was less than in Zhenyuan anti-Xa (41.9 ng/mL). This could be explained by the fact that anti-Xa assays did not directly measure the physical quantity of rivaroxaban in plasma samples, but measured the OD value of chromogenic products hydrolysed from residual FXa based on the assumption that the amount of chromogenic product was proportional to the level of factor activity in the samples[33, 34]. Therefore, anti-Xa assays may not be appropriate in scenarios requiring the detection of low concentrations, such as treatment interruption before an elective surgery, accumulation of drugs in kidney failure, judgement about medication adherence, and so on.
A few studies have reported the correlation between Biophen DiXal and HPLC-MS/MS in plasma samples [13–15, 18–20]. In general, Biophen DiXal showed a good correlation with HPLC-MS/MS, especially for concentrations < 200 ng/mL. The overall bias for Biophen DiXal was at 15.1 ng/mL, which was within the range reported in previous studies (-11.3 to 28.7 ng/mL) [13–15, 18–20]. Due to the limited sample size, the majority of previous studies could only assess correlation in samples with concentrations < 200 ng/mL, or only included several samples with high levels of concentration. Our study included 46 samples > 200 ng/mL and demonstrated that the anti-Xa assays may not be accurate enough for high-level samples.
Zhenyuan anti-Xa has been approved for several years by the NMPA; however, its correlation with HPLC-MS/MS has not yet been reported. Only a few studies have assessed its precision, accuracy, and linearity in healthy volunteers [35], or measured peak and trough concentrations in NVAF patients [36]. Our study demonstrated that Zhenyuan anti-Xa may be used to measure concentrations between 50–200 ng/mL. It should be recognised that Zhenyuan anti-Xa was only approved for the quantitative analysis of low molecular weight heparin and unfractionated heparin in clinical settings. Even though we used the rivaroxaban-specific control and calibrator samples, the accuracy and sensitivity of Zhenyuan anti-Xa were still inferior to Biophen DiXal overall, which indicates need for further optimisation of this assay.
In clinical practice, trough samples are frequently used to assess the degree of anticoagulation for patients in non-emergent situations [5]. In the subgroup analysis of trough concentrations at different doses, we found that Biophen DiXal exhibited better accuracy and precision than Zhenyuan anti-Xa. The application of Zhenyuan anti-Xa in the measurement of trough concentration may be limited, especially for patients taking rivaroxaban lower than 10 mg qd. It should be noted that a trough concentration of 0 ng/mL by Zhenyuan anti-Xa does not exclude the absence of rivaroxaban.
Our study has several limitations. First, we did not compare the correlation of the Biophen DiXal standard calibrator and low calibrator in the same sample. The Biophen DiXal Low calibrator was only used for samples < 80 ng/mL, and other samples were measured with a standard calibrator. Second, we only included 46 samples > 200 ng/mL, which may limit the interpretation of the correlation at high concentrations. Further studies may focus on the comparison of samples with high concentrations, which are common in clinical situations, such as major bleeding, overdose, and accumulation of rivaroxaban in renal failure [5–7]. Third, the lack of data about clinical outcome prevented us from analyzing dose-response relationship in Chinese patients. The results could only focus on the concentration level till now.