Keywords direct oral anticoagulants - pharmacokinetics - emergency - major bleeding - urgent surgery
Introduction
Patients with nonvalvular atrial fibrillation or venous thromboembolism require therapeutic dose anticoagulation with either vitamin K antagonists (VKA) or direct oral anticoagulants (DOAC).[1 ]
[2 ]
[3 ]
[4 ]
[5 ] With the increase in the aging population worldwide, there has been an increase in the prevalence of atrial fibrillation and venous thromboembolism and DOAC have replaced VKA in international guidelines as first choice of anticoagulant for these indications.[6 ]
[7 ] The risk of intracranial hemorrhage is lower with DOAC compared with VKA, but major bleedings still occur in 1 to 3% of patients per year.[8 ]
[9 ] The management of these bleeding complications and of urgent surgery in frail elderly patients under DOAC has become an unmet need in clinical practice and further improvement of care is warranted.[9 ]
[10 ] However, data on drug levels and pharmacokinetic (pk) of DOAC in these urgent situations are sparse.
Due to the relatively short half-life of DOAC, the timing of the last DOAC dose and the DOAC concentration at the time when the emergency event occurs is important to decide about the use of reversal agents.[11 ]
[12 ]
[13 ] Peak concentrations of DOAC are reached within 2 to 4 hours after oral intake. The half-lives of these drugs, depending on renal and hepatic function, are usually around 10 to 12 hours.[7 ] Drug levels can be determined by drug calibrated anti-factor Xa assays, or the diluted thrombin time (in case of dabigatran), while liquid chromatography-tandem mass spectrometry (LC-MS/MS) is considered the gold standard method for the measurement of DOAC levels.[14 ]
We have initiated the RADOA-registry (R eversal A gent use in patients treated with D irect O ral A nticoagulants or vitamin K antagonists), to prospectively assess outcomes in consecutive patients treated with either DOAC or VKA and admitted with major bleeding or with an indication for urgent surgery as recently described.[10 ]
[15 ] We now report on anticoagulant intensity at first presentation and drug half-life (t
1/2 ) in the DOAC-treated patients of the RADOA-registry.[10 ]
Methods
Study Design and Oversight
The RADOA-registry is a prospective, observational, noninterventional, open-label, investigator-initiated, multicenter clinical registry in Germany documenting the management of severe bleeding and/or urgent interventions in patients under treatment with VKA or DOAC. The rationale and design of the registry have been described previously.[10 ]
Patients were recruited until the predefined sample size was reached in each group. Patients were then followed prospectively until day 30 after hospital admission.
Participating centers were hospitals with 24-hour interdisciplinary teams to manage anticoagulant-related bleeding in specialized units (i.e., emergency departments and intensive care units). The study protocol was approved by all relevant institutional review boards. An external independent monitor performed 100% of the onsite source data verification.
Patients
The inclusion criteria were:
Age ≥ 18 years.
Patients anticoagulated with DOAC or VKA with clinically overt major bleeding according to a modified definition according to the International Society of Thrombosis and Haemostasis for nonsurgical patients[16 ] that presented with at least one of the following criteria: symptomatic bleeding in a critical area or organ, such as intracranial, intraspinal, intraocular, retroperitoneal, intra-articular or pericardial, or intramuscular with compartment syndrome or acute life-threatening blood loss leading to hemodynamic instability and/or acute transfusion of two or more units of whole blood or red cells.
Patients anticoagulated with DOAC or VKA needing an urgent surgical intervention within 24 hours after admission.
The registry started enrolling patients in 2014. The majority of patients was recruited between January 2016 and March 2018.
Ethics
Due to the emergency nature of the conditions under investigation, patient information and informed consent should not interfere with or delay acute treatment. With the approval of all ethics committees and institutional review boards, written informed consent was obtained from patients after the acute management phase. In the event of a patient's inability to provide written informed consent, this was obtained from his/her legal representative. Data of patients who remained unconscious or died before a legal representative had been appointed were also included. This was explicitly approved by the ethical boards to prevent major bias caused by exclusion of the most severely affected patients.[15 ] The study complies with the Declaration of Helsinki.
Substudy of the RADOA-Registry to analyze Pharmacokinetics of DOAC
In a subgroup of patients included in the RADOA-registry left-over from routine blood samples (“retention blood samples”) which were taken during the management of the acute events were collected to analyze drug concentrations and pks of DOAC during these emergency situations[10 ] (for additional information concerning recruitment see [Supplementary Fig. S1 ], available in the online version).
No additional blood sampling for pks was allowed due to the observational character of the registry. Therefore, time points at which these samples were taken were not prespecified and, thus, nonsystematic. Residual citrated plasma samples as well as serum samples were immediately frozen and stored at the participating centers at –20 or –80°C and later shipped on dry ice to the Institute for Laboratory and Transfusion Medicine, Heart and Diabetes Centre, Ruhr University Bochum, Bad Oeynhausen, Germany, where the ultraperformance liquid chromatography (UPLC)-MS/MS analysis of DOAC concentrations was centrally performed.
Sample Preparation and UPLC-MS/MS Analysis of DOACs
Sample preparation and measurement of dabigatran, rivaroxaban, apixaban, and edoxaban was performed as previously described.[17 ] In the majority of cases the measurements were performed in citrated plasma. Predilutions were automatically considered. In case of using serum or lithium heparin samples the concentrations were taken as measured.
In brief, analysis by UPLC-MS/MS was done on a two-dimensional (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) 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 DOAC of the UPLC-MS/MS method was < 0.2 ng/mL. Since DOAC values < 9 ng/mL were not clinically relevant, all time points below this value were considered zero.[17 ]
Statistical Analysis
The statistical analysis focuses on descriptive statistics (median and range or frequencies where appropriate) and two-sided 95% confidence intervals (CIs). Exponential decay of pks is assessed with linear mixed effect regression models to analyze t
1/2 together with 95% CIs and to assess associations between the exponential decay and baseline levels. As sensitivity analysis, an analogous but weighted mixed effect regression model was used which gave all patients equal weight to avoid too strong emphasis on patients with a comparable number of observations.
All statistical tests are two-sided and use a significance level of α = 5% without significance correction for multiple tests.
Results
A total of 140 patients treated with DOAC were prospectively included in the RADOA-registry. Residual plasma samples and/or serum samples for measurements of DOAC concentrations were available in 132 patients (94%) which were further analyzed. The following results only refer to this subgroup of patients.
Patient Characteristics
For pk analysis a total of 605 blood samples was available. Note that 47% (62/132) of patients were treated with apixaban, 42% (55/132) with rivaroxaban, 6% (8/132) with dabigatran, and 5% (7/132) with edoxaban. One further patient received DOAC treatment (apixaban) as well as VKA treatment (phenprocoumon) because of a medication error and was excluded from the analysis.
Of the evaluable 132 patients, 89 (67%) experienced a life-threatening bleeding event and 43 patients (33%) needed urgent surgery (not driven by severe bleeding) within 24 hours after admission.
Baseline characteristics of patients are given in [Table 1 ].
Table 1
Baseline characteristics and mortality in patients by treatment and event
Variable
Total (n = 132)
Major bleeding (n = 89)
Urgent surgery (n = 43)
Male sex, n (%)
67 (51)
47 (53)
20 (47)
Age median (1st–3rd quartile)
79 (72–84)
80 (74–84)
77 (69–83)
BMI (kg/m2 ) median (1st–3rd quartile)
26 (24–30)
26 (23–30)
27 (24–30)
Type of bleeding[a ]
Intracranial/intraspinal, n (%)
38 (29)
38 (43)
GI bleeding, n (%)
31 (24)
31 (35)
Other locations, n (%)
19 (14)
19 (21)
Type of surgery[a ]
Trauma, n (%)
23 (17)
23 (54)
Acute abdomen, n (%)
12 (9)
12 (28)
Other surgery, n (%)
12 (9)
8 (19)
Primary endpoint
30-d in-hospital mortality, n (%)
12 (9)
7 (7)
5 (12)
Abbreviations: BMI, body mass index; GI, gastrointestinal.
a Multiple types of bleeding location and surgery are possible.
Patients were on average 79 years old. In patients suffering from life-threatening bleeding 43% (38/89) presented with intracranial or intraspinal hemorrhage and 35% (31/89) had gastrointestinal bleeding. In patients with urgent surgery, 54% (23/43) had a trauma, mainly fractures and 28% (12/43) needed the intervention because of an acute abdomen. Overall, 9% (12/132) of patients died during the first 30 days after hospital admission.
Rivaroxaban Pharmacokinetics
In 52 patients treated with rivaroxaban 266 concentration measurements were performed. Time since last intake was available in 64% of rivaroxaban-treated patients and was 7.2 hours (median; 1st to 3rd quartile: 4.4–16.8 hours). Baseline samples (defined as blood sample taken within 3 hours after presentation) were not available in 13 patients. In the remaining patients rivaroxaban levels showed a high variation from undetectable levels below 9 to 803 ng/mL. A total of 206 rivaroxaban blood samples were available for pk analyses. This analysis included all samples taken during the first 3.5 days after admission ([Fig. 1 ]) but excluded patients who were re-exposed to rivaroxaban (n = 3). Fitting revealed a mean decay rate of 0.96 per day corresponding to a half-life time of 17.3 hours (95% CI: 15.4–19.7). There was no significant difference in the decay rate of the baseline levels between patients with severe bleedings and patients requiring urgent surgery (major bleeding: t
½ 16.7 hours; 95% CI: 14.7–19.3; urgent surgery: t
½ 19.7 hours; 95% CI 15.2–27.9, p = 0.292; [Fig. 1 ]). This result was overall confirmed by means of a sensitivity analysis which resulted in a slightly lower half-life time (for more information see the “Results” section of the additional statistical analysis in the [Supplementary Material ], available in the online version).
Results of the rivaroxaban levels, creatinine levels, Cockcroft–Gault formula, and coagulation assays during the time course of the registry are shown in [Table 2 ]. At baseline the median rivaroxaban level was 205 ng/mL, 59% (23/37) of patients had rivaroxaban levels > 200 ng/L and 19% (7/37) presented with rivaroxaban concentrations ≤ 75 ng/mL.
Fig. 1 Results of rivaroxaban concentrations within the first 3.5 days after admission in patients without early re-exposure to rivaroxaban. Patients reaching the 30-day in-hospital mortality are marked by red lines. Thick lines show the line corresponding to the exponential decay from a model that includes the effects and interactions of the treatment groups. Vertical axes are scaled logarithmically. Patients with major bleeding: n = 36, 161 rivaroxaban concentration measurements; Patients with urgent surgery: n = 13, 45 rivaroxaban concentration measurements.
Table 2
Rivaroxaban levels and laboratory results (number and rates, median, and 1st to 3rd quartile) in 47 patients without re-exposure to rivaroxaban within the first week after admission
Rivaroxaban-treated patients (n = 47)
Baseline
After 24 h
After 2–3 d
After 4–6 d
Rivaroxaban level ≤ 9 ng/mL[a ]
2/37 (5%)
0/29 (0%)
47/79 (59%)
31/40 (78%)
Rivaroxaban level > 9 and ≤ 30 ng/mL
0/37 (0%)
16/29 (55%)
20/79 (25%)
5/40 (13%)
Rivaroxaban level > 30 and ≤ 75 ng/mL
5/37 (14%)
8/29 (28%)
8/79 (10%)
2/40 (5%)
Rivaroxaban level > 75 and ≤ 200 ng/mL
9/37 (23%)
5/29 (17%)
4/79 (5%)
2/40 (5%)
Rivaroxaban level > 200 ng/mL
23/37 (59%)
0/29 (0%)
0/79 (0%)
0/40 (0%)
Rivaroxaban level (ng/mL)
205 (102–365)
23 (16–51)
≤ 9 (≤ 9–13)
≤ 9 (≤ 9–≤ 9)
Creatinine (mg/dL)
0.9 (0.7–1.3)
1.0 (0.8–1.6)
0.9 (0.7–1.4)
0.8 (0.6–1.2)
Cockcroft–Gault formula (mL/min)
66 (41–98)
65 (44–101)
75 (49–133)
63 (51–127)
INR[b ]
1.4 (1.3–2.0)
1.2 (1.1–1.5)
1.1 (1.1–1.2)
1.1 (1.1–1.2)
aPTT (s)[b ]
32 (28–36)
31 (28–34)
31 (28–35)
32 (28–42)
Abbreviations: aPTT, activated partial thromboplastin time; INR, international normalized ratio; PT, prothrombin time.
a Relative to all measurements. In some patients, multiple quantifications were available in the time period analyzed.
b Different PT- and aPTT-reagents were used in the participating centers (for more information see [Supplementary Table S1 ]).
Apixaban Pharmacokinetics
In 61 patients treated with apixaban 267 concentration measurements were performed. Time since last intake was available in 61% of the apixaban-treated patients and was 9.9 hours (median: 1st to 3rd quartile: 7.1–14.6 hours). Baseline samples (defined as blood sample taken within 3 hours after presentation) were not available in 15 patients. In the remaining patients apixaban concentrations showed a high variation from undetectable levels below 9 to 1,222 ng/mL.
A total of 221 apixaban blood samples were available for pk analyses. This analysis included all samples taken during the first 3.5 days after admission ([Fig. 2 ]) but excluded patients who were re-exposed to apixaban (n = 6).
Fig. 2 Results of apixaban concentrations within the first 3.5 days after admission in patients without early re-exposure to apixaban. Patients reaching the 30-day in-hospital mortality are marked by red lines. Thick lines show the line corresponding to the exponential decay from a model that includes the effects and interactions of the treatment groups. Vertical axes are scaled logarithmically. Patients with major bleeding: n = 38, 113 apixaban concentration measurements; Patients with urgent surgery: n = 22, 108 apixaban concentration measurements.
Fitting revealed a mean decay rate of 0.66 per day corresponding to a half-life time of 25 hours (95% CI: 22.9–27.6). There was a significant difference in the decay rate in patients with severe bleedings (t
½ : 20.8 hours; 95% CI: 18.8–23.2) compared with patients with urgent surgery (t
½ : 30.8 hours; 95% CI: 26.9–36.4; p < 0.001; [Fig. 2 ]). This result was overall confirmed by means of a sensitivity analysis which resulted in a slightly lower half-life time (for more information see the “Results” section of the additional statistical analysis in the [Supplementary Material ], available in the online version).
Results of the apixaban levels, creatinine levels, Cockcroft–Gault formula, and coagulation assays during the time course of the registry are shown in [Table 3 ]. On admission the median apixaban level was 108 ng/mL, 20% (9/46) of patients had apixaban levels > 200 ng/mL and 28% (13/46) presented with apixaban concentrations ≤ 75ng/mL.
Table 3
Apixaban levels and laboratory results (number and rates, median, and 1st to 3rd quartile) in 56 patients without re-exposure to apixaban within the first week after admission
Apixaban-treated patients (n = 56)
Baseline
After 24 h
After 2–3 d
After 4–6 d
Apixaban level ≤ 9 ng/mL[a ]
3/46 (7%)
1/23 (4%)
18/75 (15.0%)
24/40 (60%)
Apixaban level > 9 and ≤ 30 ng/mL
3/46 (7%)
10/23 (43%)
38/75 (25%)
10/40 (25%)
Apixaban level > 30 and ≤ 75 ng/mL
7/46 (15%)
9/23 (39%)
11/75 (10%)
6/40 (15%)
Apixaban level > 75 and ≤ 200 ng/mL
24/46 (52%)
3/23 (13%)
7/75 (9%)
0/40 (0%)
Apixaban level > 200 ng/mL
9/46 (20%)
0/23 (0%)
1/75 (1%)
0/40 (0%)
Apixaban level (ng/mL)
108 (67–181)
32 (21–58)
17 (10–30)
≤ 9 (≤ 9–15)
Creatinine (mg/dL)
1.1 (0.8–1.5)
1.0 (0.8–1.5)
1.0 (0.7–1.6)
0.9 (0.7–1.7)
Cockcroft–Gault formula (mL/min)
50 (32–79)
60 (40–91)
57 (32–85)
54 (32–76)
INR[b ]
1.3 (1.1–1.4)
1.2 (1.1–1.5)
1.2 (1.1–1.4)
1.1 (1.0–1.2)
aPTT (s)[b ]
29 (26–34)
30 (27–34)
31 (28–36)
30 (27–40)
Abbreviations: aPTT, activated partial thromboplastin time; INR, international normalized ratio; PT, prothrombin time.
a Relative to all measurements. In some patients, multiple quantifications were available in the time period analyzed.
b Different PT- and aPTT-reagents were used in the participating centers (for more information see [Supplementary Table S1 ]).
Dabigatran and Edoxaban Pharmacokinetics
In 8 patients treated with dabigatran 48 concentration measurements were performed. In 7 patients treated with edoxaban 24 blood samples for concentration measurements were available. Time since last DOAC intake was 21.5 hours in edoxaban-treated patients (median: 1st to 3rd quartile: 13.5–28.5 hours) and 7.6 hours in dabigatran-treated patients (median: 1st to 3rd quartile: 6.4–10.6 hours). Due to the small sample size statistical analysis of t ½ was not performed. Results are shown in Appendix B, [Supplementary Figs. S2 ] and [S3 ] (available in the online version).
Discussion
Baseline Anticoagulant Concentrations on Admission
In our registry, baseline DOAC concentrations on admission varied widely and ranged from 0 to more than 1,000 ng/mL, which is in agreement with findings from an observational prospective cohort study in which baseline plasma DOAC concentrations were measured in 62% of 732 DOAC-treated patients with severe bleeding.[18 ]
In our study baseline median concentrations of rivaroxaban were higher compared with apixaban concentrations due to the once daily intake of a high dose of rivaroxaban compared with apixaban which is applied at lower dosages twice daily.
Median baseline prothrombin time and activated partial thromboplastin time were only slightly prolonged in the DOAC-treated patients. These routine coagulation assays are not sensitive enough to detect clinically relevant residual DOAC concentrations on admission.[10 ]
Pharmacokinetics of DOAC in Emergency Situations
To the best of our knowledge this substudy of the RADOA-registry is the first to systematically analyze pks and half-life times in emergency situations under DOAC therapy. The observed t
½ of rivaroxaban (17.3 hours) and apixaban (25 hours) in this elderly patient population clearly exceeds the reported t
½ of these DOAC in clinical routine (apixaban: 12 hours, rivaroxaban: 11–13 hours[19 ]) and are in agreement with data of Viktil et al.[20 ] In this study evaluating 8 patients with acute hip fractures treated with DOAC, the average elimination half-live was prolonged to 21.6 hours.[20 ]
Reasons for these significantly prolonged half-lives could be altered pks due to the fragile old patient population and the emergency situation itself. As observed in patients with bleeding events in phase III trials,[1 ]
[2 ]
[3 ]
[5 ] the median age of patients in our registry was 79 years.
At baseline the median Cockcroft –Gault formula was lower in apixaban-treated patients (50 mL/min) compared with rivaroxaban-treated patients (66 mL/min), which might explain the longer t
½ of apixaban in our patient population.
Apixaban is cleared via a variety of pathways, including metabolism, biliary excretion, and direct intestinal excretion, with approximately 27% of total apixaban clearance occurring via renal excretion. Rivaroxaban is eliminated either renally (66% in total; 36% unchanged mainly through active renal secretion) or hepatobiliary.[21 ] Both drugs have a high degree of plasma protein binding of around 87% for apixaban and 95% for rivaroxaban. The volume of distribution is 21 L for apixaban and 50 L for rivaroxaban.[22 ]
[23 ] Redistribution from extravascular compartments may therefore be an additional reason for the prolonged DOAC half-lives in our patient population although the patient numbers of our registry are too small to draw definite conclusions.
The results of our study are in contrast to the results of the PAUSE study, which demonstrated that the use of a standardized protocol to stop DOAC treatment before elective procedures resulted in residual DOAC levels below 30 ng/mL in more than 67% of patients and less than 5.3% were above 50 ng/mL.[24 ] In contrast, the wide range and high levels of DOAC concentrations observed in our registry are due to the life-threatening clinical situations in which the anticoagulants cannot be stopped in time. In these critical situations rapid and quantitative determination of DOAC concentrations seems to be essential to estimate the bleeding risk in these special patient populations. Thus, all laboratories of hospitals caring for critically ill patients should be able to perform quantitative DOAC measurements in a timely manner, as recommended in the updated International Council for Standardization in Haematology (ICSH) laboratory guidelines for DOAC measurements.[25 ]
Limitations
We acknowledge that the RADOA-registry has a nonrandomized observational design. To obtain the highest possible data quality and to minimize any bias, however, patients were included prospectively and consecutively, and all enrolled patients were onsite monitored by an independent external monitor. Since we analyzed all residual blood samples available and were allowed to include patients who were unable to provide informed consent, there is minimal selection bias.
Conclusion
In conclusion, this subgroup analysis of the RADOA-registry shows that baseline concentrations of DOAC differ widely in patients admitted to hospital because of major bleeding or urgent surgery. Quantitative DOAC point of care testing on admission would improve the management of this patient group because only 70 to 80% of these patients might need specific antidotes on admission due to increased DOAC concentrations. Without rapid DOAC measurements many patients will either receive reversal agents unnecessarily which might increase thrombotic complications in this fragile patient population or may proceed to urgent treatments with high DOAC plasma levels and prolonged anticoagulant activity resulting in prolonged blood loss.
Taken together, these observations suggest the urgent need for quantitative, rapid DOAC measurement availability in all clinical laboratories as recommended by the updated ICSH DOAC laboratory guidance document,[25 ] which would be the safer alternative to the just “wait” approach in critically ill patients in these life-threatening situations. Further prospective multicenter studies are necessary to investigate the concentrations of DOACs in the perioperative urgent setting and in major bleeding in higher numbers of patients to be able to correlate these drug concentrations to clinical outcomes and to improve patient care.
What is known about this topic?
DOAC have replaced VKA in international guidelines as first choice of anticoagulant for nonvalvular atrial fibrillation and venous thromboembolism. The management of bleeding complications and of urgent surgery in patients under DOAC is a challenge.
So far data on drug levels and pharmacokinetics of DOAC in these urgent situations are sparse.
What does this paper add?
