論文ID: CR-22-0076
Background: Off-label dosing of direct oral anticoagulants (DOAC) as a treatment for non-valvular atrial fibrillation (NVAF) is problematic. Here, we investigated the status of rivaroxaban and edoxaban dosing by monitoring plasma concentrations (PCs).
Methods and Results: We monitored drug PCs in 391 and 333 outpatients receiving rivaroxaban and edoxaban, respectively, for NVAF. Drug doses were adjusted if the PC was above the cut-off value (rivaroxaban: 404 ng/mL; edoxaban: 402 ng/mL), determined from receiver operating characteristic curves for predicting bleeding events. On-label standard dosing was reduced to off-label underdosing due to high PCs above the cut-off more often for rivaroxaban (28.1%) than edoxaban (12.6%; P<0.001). Over a median follow-up of 13 months for rivaroxaban and 10 months for edoxaban, the annual incidence of bleeding events was higher with rivaroxaban than with edoxaban (4.88 vs. 3.73 patient-years; P<0.05), although no thromboembolic events occurred in either group. Furthermore, for patients with creatinine clearance >50 mL/min and body weight ≤60 kg, there was a greater incidence of bleeding events with rivaroxaban on-label 15 mg dosing than with edoxaban on-label 30 mg dosing (22.2% vs 2.9%; P<0.01).
Conclusions: Monitoring the PCs of rivaroxaban and edoxaban in NVAF patients enables dose adjustments to reduce bleeding risk. The incidence of bleeding under drug PC monitoring was less in the edoxaban than rivaroxaban group.
The 2021 European Heart Rhythm Association (EHRA) practical guide on the use of direct oral anticoagulants (DOACs) in patients with non-valvular atrial fibrillation (NVAF) stated that DOACs provide a similar degree of antithrombotic effect as vitamin K antagonists with fewer bleeding events.1 Nevertheless, in routine practice, ischemic stroke or systemic embolic events and the bleeding incidence rate are lower with apixaban and edoxaban than rivaroxaban.2–7 In addition, the EHRA practical guide suggests that special care is needed for patients with high and low body weight (BW), particularly individuals with a BW ≤60 kg, who need dose reduction in apixaban and edoxaban, but not in rivaroxaban.1
Although in actual clinical practice the use of reduced off-label doses is a problem,8 Yu et al reported that off-label overdosing of DOACs was associated with worse clinical outcomes in Asian patients than off-label underdosing.9 DOAC use reported that clinical outcomes were not worse for the off-label low-dose use than for the standard-dose use.10 These conflicting data may be attributable to the lack of anticoagulant activity monitoring. In previous studies, we found that the off-label dosing of DOACs whose suitability was proven via indirectly measuring anti-Factor Xa (FXa) activity using a chromogenic assay prevented bleeding events and avoided unsuitable dose reduction.11,12
The aims of the present study were as follows. First, we wanted to confirm which of rivaroxaban and edoxaban is the more effective and safer drug. This was achieved by identifying established indices, and their cut-off levels, for predicting bleeding events and then evaluating the appropriateness of on-label dosing using these indices and the suitability of off-label dosing using the cut-off values (or switching between these DOACs) to avoid bleeding events. Second, we re-examined the appropriateness of the dose adjustment criteria for these 2 DOACs using BW thresholds. Third, we determined the relationship between plasma concentrations (PCs) and prothrombin time (PT), a conventional index of coagulation activity (Figure 1). Throughout this trial, off-label dosing of rivaroxaban was continued since the previous our trial.
Flowchart of patient enrollment. BW, body weight; PC, plasma concentration.
After completing the previous study in June 2016, we continued this observational study at Hokusetsu General Hospital until September 2021. This study included all 724 outpatients with NVAF (433 men, 291 women; age range 40–93 years) who were undergoing treatment with rivaroxaban or edoxaban and was performed as a preliminary investigation for the multicenter study registered with the University Hospital Medical Information Network (UMIN) Clinical Trials Registry (Efficacy and Safety Evaluation According to Plasma Concentration Measurement on Direct Oral Anticoagulation Using Factor-Xa Inhibitors for Nonvalvular Atrial Fibrillation: A Multicenter Study; ID: UMIN000036769).
We calculated creatinine clearance (CrCl) using the Cockcroft-Gault formula adjusted for age, BW, serum creatinine concentration, and sex. In addition, we calculated CHADS2 scores immediately before the initiation of DOAC treatment based on the presence of heart failure, hypertension, age ≥75 years, and stroke history.
Although the Japanese standard on-label dose for rivaroxaban is 15 mg once daily (o.d.), the recommended reduced on-label dose of 10 mg o.d. rivaroxaban is used for patients with CrCl <50 mL/min, based on pharmacokinetics data.13 The standard on-label dose of edoxaban is 60 mg o.d., with on-label 30 mg o.d. being the recommended reduced dose for patients with BW ≤60 kg, CrCl <50 mL/min, or concomitantly using of P-glycoprotein inhibitors. Therefore, we selected DOAC doses for most patients based on the abovementioned dose recommendations. However, some patients started treatment with off-label reduced doses because of bleeding concerns.
All patients provided written informed consent, and this study was approved by the Ethical Review Committee at Hokusetsu General Hospital.
Drug PCsSteady state PCs were measured at least 4 weeks after the initiation of rivaroxaban or edoxaban treatment via the anti-FXa-derived indirect method using the Biophen® DiXaI kit (Hyphen Biomed, Neuville Sur Oise, France), as described in our previous report.11 This kit uses a chromogenic method based on inhibiting a constant and excess quantity of FXa by the drug being assayed, with a calibrator and control for each drug.14 Blood samples were collected 3 h after drug administration from rivaroxaban users1,11 and between 2 and 3 h after drug administration from edoxaban users.1 PCs were measured again within a few months for patients with higher or lower values in the first measurement compared with PCs in the previous DOAC study.1 To determine the reproducibility of the values obtained, we also intermittently measured PCs over a period of 6–24 months of continuous DOAC administration. If differences were observed, we used the higher values in the follow-up period. If possible, blood samples for PC measurements were also collected immediately before (trough) drug administration.
Among patients who received only on-label dosing of rivaroxaban (n=201; from October 2013 to September 2019) and edoxaban (n=166; from January 2016 to September 2019, because the DiXaI kit for edoxaban was released at the end of 2015) we sought the index related to the occurrence of bleeding events among 8 indices (age; BW; baseline CrCl; CHADS2 scores; the presence of hypertension, in particular; concomitant use of antiplatelet drugs; and trough and peak PCs). For indices found to be related to bleeding events, we determined cut-off levels for discriminating between the occurrence of bleeding or not using receiver operating characteristic (ROC) curve analysis for both drugs.
Bleeding and Thromboembolic Events in DOAC UsersAccording to the definition of the International Society on Thrombosis and Haemostasis, major bleeding is clinically overt bleeding associated with a decrease in the hemoglobin level of ≥2.0 g/dL, a transfusion requirement of ≥2 U packed red cells or whole blood, the involvement of a critical site, or a fatal outcome.15 Non-major clinically relevant bleeding is defined as clinically overt bleeding that does not meet the criteria for major bleeding, but that requires medical intervention, unscheduled consultation with a physician, or temporary discontinuation of study treatment and results in pain or impairment of daily activities; overt bleeding episodes not meeting the criteria for major or non-major clinically relevant bleeding are classified as minor bleeding.16 Thromboembolic events were defined using the same definitions as the outcome and endpoint components in the J-ROCKET AF study.16
All patients in this study were followed up in Hokusetsu General Hospital or by a general physician. Blood samples were collected from patients who experienced a bleeding event or neurological symptoms, and the next course of treatment (e.g., drug interruption or further therapeutic procedures) was determined. We also recorded follow-up outcomes by reviewing outpatient charts or via telephone calls to the general physicians and patients. All data collection stopped on September 30, 2021.
Dose Adjustments and Drug SwitchingFor patients with peak PCs above the cut-off for bleeding events under rivaroxaban and edoxaban use, we initiated off-label underdosing with suitable anticoagulability proven by measuring anti-FXa activity to prevent further bleeding events, after obtaining consent for dose adjustment from the patients and the Ethical Review Committee at Hokusetsu General Hospital. In addition, for patients with bleeding events and/or high peak PCs above each cut-off PC, we switched their respective drugs to another DOAC. We continued to monitor the incidence of events following the initiation of the new dosing or drug protocol using the same observational protocol.
Re-Evaluation of Dose Reduction Criteria With BW in RivaroxabanAmong rivaroxaban users, we further re-evaluated the peak PC data and the incidence of bleeding events with the additional index of BW (i.e., patients were divided into 2 groups according to a BW threshold of 60 kg: BW ≤60 or >60 kg).
Association Between PC and PTWe examined the utility of PT as a substitute for PCs. We evaluated the relationship between PT (using Thromborel® S; Siemens Healthcare Diagnostics, France) and PCs in both rivaroxaban and edoxaban users, and examined the threshold PT levels for bleeding using ROC curve analysis.
Statistical AnalysesStatistical analyses were performed based on a 5% level of significance. We used the Wilcoxon 2-sample test to compare continuous variables. We used simple linear regression and correlation tests to analyze the correlation coefficients, and performed logistic regression analyses to determine the risk factors related to bleeding events using Mac Statistical Analysis v. 3.0 (Esumi Company, Tokyo, Japan; http://www.esumi.co.jp/). Data are expressed as the mean±SD.
We initially evaluated the indices related to the incidence of bleeding events among patients treated with on-label rivaroxaban (n=115) and on-label edoxaban (n=85) for whom samples of both peak and trough PCs were available. Of the 8 indices, only peak PCs could predict bleeding events in patients taking rivaroxaban or edoxaban based on univariate and multivariate logistic regression analyses (Supplementary Table 1). Second, the cut-off level for bleeding events among patients with peak PC data taking on-label rivaroxaban (n=201) and on-label edoxaban (n=166) was evaluated. In this analysis, 21 and 7 bleeding events were identified in the rivaroxaban and edoxaban groups, respectively, and the peak PC cut-off values for predicting bleeding were found to be 404 ng/mL in rivaroxaban and 402 ng/mL in edoxaban (Supplementary Figure).
Patient CharacteristicsWe examined 724 patients with NVAF who were receiving treatment with rivaroxaban (n=391) or edoxaban (n=333). These patients included those using on-label standard doses, those using on-label reduced doses, and those using off-label reduced doses after adjustment of the standard dose.
Table 1 shows the profiles of these DOAC users. There was no clear difference in terms of age between rivaroxaban and edoxaban users. BW was slightly lower for those using rivaroxaban standard-dose than those using edoxaban standard-dose. Conversely, BW was lower for those using edoxaban reduced-dose than for those using rivaroxaban reduced-dose. There were no differences in terms of baseline CrCl levels, CHADS2 score, NVAF pattern, and antiplatelet drug use between the 2 DOAC groups (Table 1). Age, BW, CrCl levels, and CHADS2 scores in the off-label reduced-dose group were intermediate between the standard-dose and reduced-dose groups (Table 1).
| Rivaroxaban (n=391) | Edoxaban (n=333) | Rivaroxaban vs. edoxaban |
||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Group 1: on-label 15 mg (n=214) |
Group 2: off-label 10 mg (n=110) |
Group 3: on-label 10 mg (n=67) |
P value | Group 4: on-label 60 mg (n=83) |
Group 5: off-label 60 mg (n=5) |
Group 6: off-label 30 mg (n=42) |
Group 7: on-label 30 mg (n=203) |
P value | ||
| Age (years) | 66±9 | 70±9 | 79±6* | *P<0.01 vs. Group 1 |
69±8† | 60–84 | 75±7 | 79±8*,† | *P<0.01 vs. Group 4 |
†NS vs. Group 1 or Group 3 |
| Males/Females (n) |
160/54 | 68/42 | 31/36 | 64/19 | 1/4 | 32/10 | 77/126 | |||
| BW (kg) | 69±15 | 65±13 | 54±10* | *P<0.01 vs. Group 1 |
74±13† | 50–58 | 70±9 | 51±8*,† | *P<0.01 vs. Group 4 |
†NS: Group 1 vs. Group 4; Group 3 vs. Group 7 |
| Baseline CrCl (mL/min) |
82±29 | 71±19 | 41±9* | *P<0.01 vs. Group 1 |
80±24† | 52–64.5 | 69±18 | 44±14*,† | *P<0.01 vs. Group 4 |
†NS: Group 1 vs. Group 4; Group 3 vs. Group 7 |
| CHADS2 score | 1.5±0.8 | 1.7±0.9* | 2.1±1.0* | *P<0.01 vs. Group 1 |
1.8±0.8 | 1–2 | 2.0±1.1 | 2.2±1.1*,† | *P<0.01 vs. Group 4 |
†NS: Group 1 vs. Group 4; Group 3 vs. Group 7 |
| NVAF pattern: paroxysmal/ permanent (n) |
88/125 | 45/63 | 25/42 | 33/50 | 2/3 | 13/29 | 87/116 | |||
| No. APT use/no use (%) |
29/185 (13.6) |
15/95 (13.9) |
13/54 (19.4) |
8/75 (9.6) |
No use | 5/37 (9.5) |
36/167 (17.7) |
|||
| Follow-up duration (months) |
13 [6–32.5] |
18.5 [6–36] |
12 [5–26] |
10 [6–18] |
40 [18–50] |
12 [6–24] |
12 [5–20] |
|||
Unless indicated otherwise, data are given as the mean±SD or median [interquartile range]. Group 5: off-label overdose use of edoxaban 60 mg. APT, antiplatelet drug; BW, body weight; CrCl, creatinine clearance; NVAF, non-valvular atrial fibrillation.
On-label dosing was continued in patients with a peak PC below each cut-off level. However, the dose was reduced in patients who developed bleeding events or showed peak PCs above each cut-off level while using the on-label standard dose. For patients with peak PCs above each cut-off level, 106 were using rivaroxaban (27.1% of all users; 76 on-label 15 mg, 15 off-label 10 mg, and 15 on-label 10 mg) and 19 were using edoxaban (5.7% of all users; 15 on-label 60 mg, 1 off-label 30 mg, and 3 on-label 30 mg). Peak and trough PCs for rivaroxaban and edoxaban users are shown in Figure 2.
Distribution of trough (T) and peak plasma concentrations (PCs) in patients using (A) on-label 15 mg, off-label 10 mg, and on-label 10 mg rivaroxaban and (B) on-label 60 mg, off-label 60 mg (overdosing), off-label 30 mg, and on-label 30 mg edoxaban. Bleeding events (red circles) occurred in a total of 39 (10.0%) and 18 (5.4%) of patients using rivaroxaban and edoxaban, respectively. Red numbers and percentages in parentheses indicate the incidence of bleeding in each dosing group. In both the rivaroxaban and edoxaban groups, there were no differences in trough drug PCs between patients with and without bleeding events, except for 1 patient using off-label 60 mg edoxaban (overdose).
Rivaroxaban 10 mg was switched to edoxaban 30 or 15 mg in users who developed bleeding events. Similarly, the edoxaban dose was reduced from 30 to 15 mg in users who developed bleeding events. These dose adjustments were performed after obtaining consent for off-label use from each patient. With these dose adjustments, none of the patients experienced rebleeding or new thromboembolic events. The data on drug switching is presented in the following section; we hope to present the outcomes of edoxaban 15 mg use in another paper.
Bleeding Events With DOAC UseAmong all rivaroxaban users, 28.1% (110/391) were using off-label rivaroxaban 10 mg. In all, 39 rivaroxaban users developed bleeding events (15 major, 8 non-major, and 16 minor bleeding) (Table 2). The incidence of bleeding events was significantly lower among those using off-label rivaroxaban 10 mg (3/110; 2.7%) than among the 2 groups of on-label rivaroxaban users (13.6% and 10.4%; Figure 2A).
| Rivaroxaban (n=391) | Edoxaban (n=333) | |||
|---|---|---|---|---|
| Dosing | On-label 15 mg, n=214 | On-label 60 mg, n=83 | ||
| Off-label 10 mg, n=110 (28.1%) | Off-label 30 mg, n=42 (12.6%) | |||
| On-label 10 mg, n=67 | On-label 30 mg, n=203 | |||
| Off-label 60 mg, n=5 | ||||
| Bleeding events | Yes (n=39; 10.0%)** | No (n=352) | Yes (n=18; 5.4%) | No (n=315) |
| On-label 15/60 mg | 29/214 (13.6%) | 9/83 (10.8%) | ||
| Off-label 10/30 mg | 3/110 (2.7%) | 1/42 (2.4%) | ||
| On-label 10/30 mg | 7/67 (10.4%)** | 4/203 (2.0%) | ||
| Off-label 60 mg | – | 4/5 (80%) | ||
| Annual bleeding rate (patient-years) | 4.88 | – | 4.80/3.73* | – |
| Age (years) | 70±9 | 70±10 | 75±8 | 76±9 |
| BW (kg) | 62±11 | 65±15 | 62±11 | 59±14 |
| Baseline CrCl (mL/min) | 67±22 | 72±2 | 55±14 | 57±24 |
| CHADS2 score | 1.7±0.8 | 1.6±0.9 | 2.1±1.1 | 2.1±1.0 |
| HT (+/−) (n) | 30/9 | 265/87 | 15/3 | 228/87 |
| APT use (%) | 8 (21)† | 45 (13) | 1 (7)† | 42 (13) |
| Peak PC (ng/mL) | 459±54*** | 331±103 | 463±70*** | 241±78 |
| Trough PC (ng/mL) (no. trough samples) | 42±38 (n=27) | 41±159 (n=161) | 45±22 (n=13) | 39±22 (n=140) |
Unless indicated otherwise, data are given as the mean±SD or n (%). *P<0.05 compared with all rivaroxaban users (but excluding the 5 patients using off-label edoxaban overdosing); **P<0.01 compared with the same level edoxaban dose; ***P<0.01 compared with the no-bleeding group for each direct oral anticoagulant (DOAC); †NS compared with the no-bleeding group for each DOAC. HT, hypertension; PC, plasma concentration. Other abbreviations as in Table 1.
The proportion of off-label edoxaban reduced dose users (12.6%) was lower than that of off-label rivaroxaban users (28.1%). Edoxaban caused bleeding events in 18 patients (5 major, 6 non-major, and 7 minor bleeding). Of these patients, 4 were using off-label edoxaban 60 mg (overdosing), and 14 were using on- or off-label edoxaban reduced-dose. The incidence of bleeding events was significantly lower among those using off- and on-label edoxaban 30 mg (2.4% and 2.0%) than among those using on-label edoxaban 60 mg (10.8%; Figure 2B). In 1 patient using on-label 30 mg edoxaban with concomitant verapamil dosing, minor bleeding events occurred but ceased after verapamil withdrawal. In terms of outcomes before and after starting off-label dosing with edoxaban in September 2019, 7 of 166 patients (5 on edoxaban 60 mg and 2 on on-label edoxaban 30 mg) and 7 of 167 patients (4 on edoxaban 60 mg, 1 on off-label edoxaban 30 mg, and 2 on on-label edoxaban 30 mg) developed bleeding events. Although the incidence of bleeding events in those using edoxaban 60 mg did not significantly change after starting off-label dosing, it may have been reduced with off-label dosing. Furthermore, the 42 patients who developed bleeding events (n=18) or had peak PCs above each cut-off (n=24) underwent drug switching or dose reduction from on-label edoxaban 60 mg (n=19) or rivaroxaban 15 or 10 mg (n=23) to off-label edoxaban 30 mg.
Rivaroxaban users (4.88 patient-years; 39 events for 8 years; 10.0%) had a higher annual incidence of bleeding events than edoxaban users, except for the 4 patients in the overdosing group (3.73 patient-years; 18 events in 5.75 years; P<0.05). However, the overall incidence of bleeding was 4.80 patient-years among edoxaban users. The incidence of bleeding events among on-label reduced-dose users was significantly higher in rivaroxaban than in edoxaban (10.4% vs 2.0%; P<0.05). Patients with bleeding events had significantly higher peak PCs for both rivaroxaban and edoxaban than those without bleeding events. However, there were no differences in terms of trough PCs between the 2 groups.
Details of peak PCs and bleeding events in the 79 patients who developed these events are presented in Supplementary Table 2. Among these patients, 23 with peak PCs below each cut-off level developed bleeding events. Thus, some bleeding events were not drug induced and were related to organic issues and/or an increase in the potential activity of bleeding by DOAC dosing.
Based on these data, acceptable PCs for predicting bleeding events caused by each drug dose were defined as greater than the 5th percentile peak PC and lower than the peak PC cut-off level. Acceptable PCs were 153.4–404 ng/mL for rivaroxaban and 125.5–402 ng/mL for edoxaban. These data are similar to the expected PCs shown in the EHRA practical guide.1 Among the off-label users receiving each reduced dose, the peak PCs were under control. None of the patients experienced a thromboembolic event, particularly among patients receiving off-label reduced dosing, with the exception of some patients who experienced a thromboembolic event after drug discontinuation. Table 1 shows the median duration of follow-up under DOAC dosing in the different groups.
Drug Switching or Dose Reduction in Identical Patients With Bleeding Events and/or High PCs Caused by Rivaroxaban or EdoxabanDrug switching between rivaroxaban and edoxaban was performed in 86 identical patients. Higher peak PCs during rivaroxaban use decreased after switching to edoxaban in all except 4 patients. These 4 patients developed bleeding events with higher peak PCs after switching to edoxaban 60 mg, which resulted in off-label overdosing. However, the peak PCs in these patients were controlled with on-label edoxaban 30 mg. Figure 3 shows changes in peak PCs with these adjustments.
Changes in peak plasma concentrations (PCs) with drug switching from rivaroxaban to edoxaban (n=86) or dose reduction to an off-label reduced dose (n=58) in identical patients with bleeding events and/or high PCs under rivaroxaban (R) and edoxaban (E) treatment. Peak PCs decreased not only with drug switching, but also with dose reduction. However, 4 patients (red lines) experienced bleeding events because of the off-label of edoxaban 60 mg (overdose). The boxes show the interquartile range, with the median value indicated by the horizontal line; whiskers show the range. In each box, the “X” and numbers next to the box indicate the mean values. BW, body weight.
Dose reduction from on-label standard to off-label reduced doses was performed (Figure 2). The left side of Figure 3 shows changes in peak PCs with dose reduction in rivaroxaban from on-label standard dose to off-label reduced dose in 43 identical patients. However, in some patients with high peak PCs even after dose reduction to rivaroxaban off-label 10mg, treatment was switched to edoxaban.
Repeated measurements of peak PCs during DOAC dosing showed 3 patterns: (1) almost stable PCs over time; (2) a PC greater than the cut-off already at the first measurement; and (3) PCs exceeded the cut-off after a few months or years. Among patients with peak PCs above the cut-off for bleeding events under rivaroxaban, neither physicians nor patients preferred continuing the therapeutic course without switching after reviewing the data during the initial term on drug switching. Therefore, we could not perform a randomized comparative evaluation between those with and without dose switching or switching between 2 drugs. As mentioned in the previous section, there were no events experienced by patients, even after drug switching or off-label dosing.
Standard and Reduced Dosing From the Viewpoint of BWThe edoxaban standard dose was first determined using the dose adjustment criterion of BW >60 kg. However, the rivaroxaban standard dose was determined based on a CrCl of >50 mL/min. Therefore, using a BW of 60 kg as the threshold, we divided rivaroxaban standard dose users into 2 subgroups: BW ≤60 and >60 kg. Next, peak PCs and the incidence of bleeding were evaluated. As indicated in Table 3, 25.2% of rivaroxaban 15 mg users had a BW ≤60 kg (Group A1). A higher percentage of females had a BW ≤60 kg than a BW >60 kg (Group A2; P<0.001). In addition, patients with a BW ≤60 kg had lower CrCl (P<0.01), higher peak PCs (P<0.01), and a greater incidence rate of bleeding events (P<0.05) than those with a BW >60 kg. Further, we found no differences in these 3 indices between on-label edoxaban 60 mg users (Group B) and on-label rivaroxaban 15 mg users with a BW >60 kg (Group A2). Furthermore, among users with a BW ≤60 kg and CrCl >50 mL/min, we compared the background characteristics between patients using rivaroxaban 15 mg (Group A1) and those using edoxaban 30 mg (Group D1–2, n=69, among Group D1 [BW ≤60 kg without reference to CrCl], n=179). There were no differences in terms of the percentage of females, age, and BW between these 2 DOAC groups. However, rivaroxaban users (Group A1) had higher peak PCs (P< 0.01) and a greater incidence rate of bleeding events (P< 0.01) than edoxaban users (Group D1–2; Figure 4).
| CrCl ≥50 mL/min |
Rivaroxaban 15 mg (n=214) | Edoxaban 60 mg (n=83) | Edoxaban 30 mg | |||||
|---|---|---|---|---|---|---|---|---|
| BW ≤60 kg n=54 (25.2%) |
BW >60 kg n=160 (74.8%) |
A1 vs. A2 | BW >60 kg n=83 (100%) |
A1 vs. B | A2 vs. B | BW ≤60 kg n=69 (34.0%) |
A1 vs. D1–2 | |
| Group | A1 | A2 | B | D1–2 | ||||
| Males/females (n) | 24/30 | 136/24 | 65/18 | P<0.01 | NS | 23/46 | ||
| % Female | 55.6 | 15 | P<0.001 | 22 | 66.7 | NS | ||
| Age (years) | 68±7 | 66±10 | NS | 69±8 | NS | NS | 74±7 | NS |
| BW (kg) | 52±5 | 74±13 | P<0.01 | 74±13 | P<0.01 | NS | 53±5 | NS |
| CrCl (mL/min) | 66±15 | 87±30 | P<0.01 | 79±25 | P<0.01 | NS | 61±10 | P<0.05 |
| Peak PC (ng/mL) | 406±106 | 353±105 | P<0.01 | 328±74 | P<0.01 | NS | 220±76 | P<0.01 |
| Bleeding incidence (%) |
12/54 (22.2) | 17/160 (10.6) | P<0.05 | 9/83 (10.8) | P<0.05 | NS | 2/69 (2.9) | P<0.01 |
| CrCl 30–50 mL/min |
Rivaroxaban 10 mg (n=67) | Edoxaban 30 mg (n=203) | ||||||
| BW ≤60 kg CrCl <50 n=45 (67.2%) |
BW >60 kg CrCl <50 n=22 (32.8%) |
C1 vs. C2 | BW ≤60 kg No relation to CrCl level n=179 (88.2%) |
BW >60 kg CrCl <50 n=24 (11.8%) |
C1 vs. D1 | C2 vs. D1 | C2 vs. D2 | |
| Group | C1 | C2 | D1 | D2 | ||||
| Male/females (n) | 13/32 | 18/4 | P<0.01 | 70/109 | 15/9 | NS | P<0.01 | NS |
| % Female | 71 | 18 | 61 | 38 | ||||
| Age (years) | 80±6 | 78±6 | NS | 79±8 | 80±7 | NS | NS | NS |
| BW (kg) | 48±7 | 66±4 | P<0.01 | 50±7 | 65±4 | NS | NS | NS |
| CrCl (mL/min) | 39±10 | 43±5 | NS | 45±6 | 40±5 | NS | NS | NS |
| Peak PC (ng/mL) | 315±98 | 299±87 | NS | 224±73 | 215±87 | P<0.01 | P<0.01 | P<0.01 |
| Bleeding incidence (%) |
6/45 (13.3) | 1/22 (4.5) | NS | 3/179 (1.7) | 1/24 (4.2) | P<0.01 | NS | NS |
BW, body weight; CrCl, creatinine clearance; PC, plasma concentration.
Peak plasma concentrations (PCs) and the incidence of bleeding in patients with a body weight of ≤60 kg and a creatinine clearance of >50 mL/min using rivaroxaban (Riv) on-label 15 mg (n=54) and edoxaban (Edox) on-label 30 mg (n=69). Rivaroxaban on-label 15 mg users (Group A1 in Table 3) had higher peak PCs (box plots) and a greater incidence rate of bleeding events (red), compared with patients using on-label edoxaban 30 mg (Group D1–2 in Table 3). The boxes show the interquartile range, with the median value indicated by the horizontal line; whiskers show the range. In each box, the “X” and numbers next to the box indicate the mean values.
Among those using on-label reduced doses with CrCl 30–50 mL/min, 67 were using rivaroxaban and 203 were using edoxaban. As indicated in Table 3, edoxaban users had a lower peak PC and incidence of bleeding events than rivaroxaban users.
Association Between PT and PC and Cut-Off PT Levels for Bleeding EventsThere was a positive correlation between the PT levels measured using Thromborel® S and PCs in 280 samples collected from edoxaban users (Figure 5). However, this correlation was poorer in rivaroxaban users. Using this correlation in edoxaban users, we determined that the peak PT cut-off level to predict bleeding events was 21.7 s, with an area under the curve of 0.857. In contrast, in rivaroxaban users with a PT >20 s, cautious observation with a complete blood count is necessary to evaluate bleeding.
Relationship between prothrombin time (PT), measured using Thromborel® S, and plasma concentrations (PCs) in patients receiving rivaroxaban (Left) or edoxaban (Right). There was a good correlation between PT and PCs (r=0.883, P<0.0001) in the 280 samples from edoxaban users, but a poor correlation in rivaroxaban users. Using this correlation in edoxaban users, we determined a cut-off level of 21.7 s for peak PT (area under the curve [AUC] 0.857; Lower right) for predicting bleeding events in the 280 on-label users with PT data. CI, confidence interval.
This study was performed to re-evaluate the efficacy and the safety of edoxaban and rivaroxaban dosing, coupled with monitoring of peak and trough PCs of both DOACs using anti-FXa chromogenic assays.
The annual incidence of bleeding events was significantly higher among rivaroxaban than edoxaban users, excluding those being overdosed (4.88 vs. 3.73 patient-years; P<0.05). Because of the development of bleeding events or peak PCs above cut-off levels, the proportion of acceptable dose adjustment using off-label reduced dose with monitoring of PCs was 28.1% among all rivaroxaban users and 12.6% among all edoxaban users (P<0.001). Further, the incidence in edoxaban users was less than half of that in rivaroxaban users. Direct comparison of drug switching between rivaroxaban and edoxaban in identical patients showed that lower peak PCs were maintained in edoxaban users. Furthermore, in patients with a BW ≤60 kg, on-label edoxaban 30 mg may be safer for reducing bleeding events than on-label rivaroxaban dosing. Thus, edoxaban is associated with a lower incidence of bleeding events than rivaroxaban, although the cut-off levels for bleeding were similar between the 2 DOACs. Various previous reports have indicated that rivaroxaban use has a greater potential for the development of bleeding events.2,7,17–20
In patients with low BW, the dose adjustment criteria indicate that BW ≤60 kg requires a dose reductions in apixaban and edoxaban. The US Food and Drug Administration (FDA) practical guide on DOAC use recommended adjusted dosing of apixaban or edoxaban with laboratory monitoring to assess drug effect in patients with a low BW (<60 kg).21 Similarly, our data indicate that among patients with a BW ≤60 kg, those using rivaroxaban standard-dose developed higher peak PCs and a greater incidence of bleeding events. Thus, among on-label users with similar characteristics (i.e., CrCl >50 mL/min, BW ≤60 kg), edoxaban is safer in terms of bleeding than rivaroxaban.
Some clinical studies have reported the development of thromboembolic events with off-label underdosing.7,8,22,23 In constant, other studies reported that off-label dosing does not always cause thromboembolic events.9,10 In the present DOAC study, bleeding events were reduced by controlling drug PCs with off-label underdosing. Based on these data, the activity of DOAC should be monitored to resolve these problematic issues.1,24–26
Trough PCs have been found to be useful in monitoring anticoagulant activity in those using FXa inhibitors, although there is no comparative study between trough and peak PCs.27,28 However, the maximum effect of the DOAC on the clotting test occurs at its maximum PC.1 Therefore, in our experience, patients with high peak PCs and a greater incidence of bleeding events fall into 1 of 2 categories of trough PCs: 1 group has a high trough PC near or above 100 ng/mL and the other has trough PCs <60 ng/mL (as shown in Figures 2,3). In addition, current reports show that peak PC or anti-FXa activity is significantly related to hemorrhagic events.11,29,30 Therefore, a high trough PC may be sufficient for predicting bleeding events in a few patients. However, some patients have low trough PCs despite developing bleeding events under a high peak PC. Although trough PC is useful for predicting bleeding events, it is unsatisfactory for the evaluation of bleeding events, as shown in the present study. Thus, peak PCs are more suitable for predicting bleeding events.
PT was proposed to monitor the anticoagulant activity of DOACs in medical facilities in which a chromogenic assay is not available. We detected a good correlation between PCs and PT levels with edoxaban dosing. Although the potential of peak PT for evaluating bleeding events was inadequate, we could determine whether suitable anticoagulant activity was preserved when the peak PT was within 15–20 and <21.7 s, which is the cut-off level for bleeding events.
The present study has some limitations. It was conducted at a single institution, and the sample size was <400 patients per drug. In addition, in Japan, the approved standard dose of rivaroxaban is 15 mg and the reduced dose is 10 mg, which differ from the globally approved doses. In patients with a CrCl >50 mL/min and a BW ≤60 kg, a reduced dose (30 mg) of edoxaban and a standard dose (15 mg) of rivaroxaban are required. However, in our analysis using BW, switching from edoxaban to rivaroxaban in these patients was detrimental. Therefore, we did not switch to rivaroxaban. In the present study, the Biophen DiXaI kit was used for the chromogenic anti-FXa assay. Considering that only a few kits currently provide a chromogenic assay similar to that mentioned above, the cut-off PC for predicting bleeding events may vary depending on the quality of the kit used. Monitoring the PCs of rivaroxaban and edoxaban safely reduced their dosing without any increase in the incidence of thromboembolic events in our cohort. Although the few ischemic events in our single-center study may derive from the small sample size and short follow-up period, we assume that a dose reduction of drugs based on PCs can prevent bleeding events and drug discontinuation, which lead to subsequent thromboembolic events. Therefore, another multicenter study must be performed to evaluate whether dose reductions with confirmation of anticoagulant activity are safe and can prevent bleeding events.
Among all rivaroxaban and edoxaban users, the incidence of bleeding events and the use of off-label reduced doses was lower in edoxaban users than in half of the rivaroxaban users. In addition, in a direct comparison of drug switching in the same patients, edoxaban users had lower peak PCs than rivaroxaban users. Hence, edoxaban has a higher safety profile than rivaroxaban, although the cut-off levels for predicting bleeding events for the 2 drugs are similar.
No funding was obtained for this article.
The authors have no conflict of interests to disclose
The Ethical Review Committee at Hokusetsu General Hospital approved this study (May 20, 2016).
Please find supplementary file(s);
https://doi.org/10.1253/circrep.CR-22-0076
