A total of 806 patients were identified, of whom 193 were treated with apixaban, 15 with dabigatran, 74 with edoxaban, and 524 with rivaroxaban. Due to the very low number of patients on dabigatran and edoxaban, only data from patients on apixaban and rivaroxaban were further analysed.
Median age of the remaining 717 patients was 61 (range 20–90, IQR 52–73) years and 392 were female. There were two plasma samples from different time points available in 238 patients, making a total of 955 plasma samples (252 apixaban and 703 rivaroxaban) included into the study. Median age was 61 (IQR 52–73) years in the apixaban group and 61 (IQR 52–73) years in the rivaroxaban group. One patient was switched from rivaroxaban to apixaban. This patient was excluded in the patient’s characteristics displayed in Table 1.
Table 1
| | Apixaban | | Rivaroxaban | | |
| | n | % | n | % | p |
Anamnesis | Female gender | 108/193 | 55.9 | 284/524 | 54.2 | 0.675 |
| Family history of VTE | 43/193 | 22.4 | 144/524 | 27.5 | 0.159 |
| History of stroke | 9/193 | 4.7 | 15/524 | 2.9 | 0.234 |
| Myocardial infarction | 6/193 | 3.1 | 10/524 | 1.9 | 0.334 |
| Arterial embolism | 6/193 | 3.1 | 4/524 | 0.8 | 0.027 |
| Renal insufficiency | 16/193 | 8.3 | 47/524 | 9.0 | 0.776 |
| Obesity | 24/193 | 12.4 | 72/524 | 13.7 | 0.649 |
| History of cancer | 19/193 | 9.8 | 40/524 | 7.6 | 0.339 |
| Active cancer | 4/193 | 2.1 | 10/524 | 1.9 | 1.000 |
| Diabetes mellitus | 11/193 | 5.7 | 46/524 | 8.8 | 0.176 |
| Smoking | 25/193 | 13.0 | 68/524 | 13.0 | 0.993 |
| Oral contraceptive | 4/193 | 2.1 | 5/524 | 1.0 | 0.260 |
Thrombophilia | Antithrombin deficiency | 4/193 | 2.1 | 10/524 | 1.9 | 1.000 |
| Protein C deficiency | 2/193 | 1.0 | 7/524 | 1.3 | 1.000 |
| Protein S deficiency | 5/193 | 2.6 | 9/524 | 1.9 | 0.541 |
| Factor V-Leiden mutation | 34/193 | 17.6 | 113/524 | 21.6 | 0.281 |
| Prothrombin-G20210A-Mutation | 12/193 | 6.2 | 39/524 | 7.4 | 0.447 |
| Combined thrombophilia | 3/193 | 1.6 | 13/524 | 2.5 | 0.578 |
Reason for anticoagulation | Chronic DVT | 139/193 | 72.0 | 440/524 | 84.0 | < 0.001 |
| Recurrent DVT | 27/193 | 14.0 | 94/524 | 17.9 | 0.210 |
| Pulmonary embolism | 97/193 | 50.3 | 218/524 | 41.6 | 0.038 |
| Recurrent pulmonary embolism | 7/193 | 3.6 | 24/524 | 4.6 | 0.578 |
| Recurrent VTE | 34/193 | 17.6 | 110/524 | 21.0 | 0.317 |
| Atrial fibrillation | 23/193 | 11.9 | 30/524 | 5.7 | 0.005 |
Reference cohort
For the determination of reference values, plasma samples from a cohort of 50 healthy subjects coming to the outpatient coagulation center MVZ Limbach Magdeburg, Germany for thrombophilia testing because of a family history of hereditary thrombophilia were analyzed. These subjects were not on anticoagulation, had no thrombophilia and no history of thromboembolism.
Median age of the reference population was 62 (IQR 56–66) years and 25 (50.0%) were females. Mean (2.5th -97.5th percentile) TG parameters of this cohort were: lag time 4.5 (3.0-6.6) min; TTP 9.8 (7.4–13.1) min; peak thrombin 314 (177–465) nmol/L; velocity index 63 (22–123) nmol/min; ETP 3207 (1946–3874) nmol/L*min.
Correlation between DOAC plasma levels and thrombin generation parameters
There was a significant correlation between DOAC plasma levels and thrombin generation parameters except for the lag time with apixaban. The correlation between DOAC plasma levels and the TG parameters TTP, peak thrombin and ETP in all included patients is shown in Fig. 1.
The correlation beween DOAC plasma levels and TG parameters was higher for rivaroxaban compared to apixaban in patients with AF, patients with thrombophilia and in the subgroup of thrombophilia patients with Factor-V-Leiden-Mutation and Prothrombin-G20210A-Mutation. In addition, there was a higher correlation between DOAC plasma levels and TG parameters in patients with thrombophilia, especially in patients with Prothrombin-G20210A-Mutation, compared to patients with AF (supplementary table 3).
Comparison of TG parameters at different drug levels
For the comparison of TG parameters at different drug concentrations, drug levels were divided into the subgroups < 20 ng/mL, 21–50 ng/mL, 51–100 ng/mL, 101–200 ng/mL, 201–300 ng/mL and > 300 ng/mL. Patients treated with rivaroxaban showed a significant increase in the TG parameters lag time and TTP and a drop in peak thrombin, velocity index and ETP between all subsequent drug level categories except for the highest drug levels. Patients on apixaban showed these differences only between the drug levels 21–50 ng/mL and 51–100 ng/mL.
Rivaroxaban-treated patients had a longer TTP compared to patients on apixaban at all plasma levels > 20 ng/mL. Peak thrombin, velocity index and ETP were lower and lag time longer in patients on rivaroxaban at drug levels between 100 and 300 ng/mL. The comparison of thrombin generation between rivaroxaban and apixaban is given in Table 2 and the p-values for the comparison of the TG parameters between different drug levels in the supplementary table 4.
Table 2
Thrombin generation parameters in apixaban and rivaroxaban treated patients at different drug levels. Values are given as median and interquartile range in brackets.
| | n | Drug level | Lag time | TTP, min | Peak thrombin | VI | ETP |
< 20 ng/mL | Apixaban | 41 | 0 (0–13) | 4.6 (3.8–5.2) | 9.3 (8.2–10.8) | 360 (256–460) | 70 (42–124) | 2896 (2633–3511) |
Rivaroxaban | 186 | 10 (0–14) | 4.7 (4.0-5.6) | 9.6 (7.9–11.6) | 324 (238–469) | 69 (37–132) | 3122 (2645–3476) |
p | | 0.310 | 0.397 | 0.475 | 0.807 | 0.793 | 0.518 |
21–50 ng/mL | Apixaban | 27 | 39 (32–45) | 4.5 (3.9–5.4) | 9.8 (8.6–11.4) | 279 (192–348) | 55 (31–86) | 2961 (2648–3332) |
Rivaroxaban | 93 | 33 (27–42) | 5.0 (4.3–5.8) | 11.1 (9.4–13.3) | 224 (170–360) | 36 (22–82) | 2745 (2415–3121) |
p | | 0.005 | 0.140 | 0.018 | 0.194 | 0.085 | 0.048 |
51–100 ng/mL | Apixaban | 81 | 71 (61–83) | 5.3 (4.1–6.2) | 11.9 (10.2–13.5) | 164 (114–214) | 26 (16–38) | 2285 (1660–2796) |
Rivaroxaban | 72 | 73 (61–85) | 5.4 (4.4–6.6) | 13.5 (11.4–15.8) | 172 (123–219) | 22 (13–34) | 2402 (1980–2757) |
p | | 0.946 | 0.437 | < 0.001 | 0.473 | 0.207 | 0.208 |
101–200 ng/mL | Apixaban | 69 | 140 (117–163) | 5.1 (4.1–6.1) | 11.2 (9.6–13.0) | 170 (108–232) | 29 (16–48) | 2247 (1621–2892) |
Rivaroxaban | 98 | 153 (123–178) | 6.3 (5.5–7.4) | 16.2 (14.0-19.6) | 123 (86–164) | 12 (7–18) | 1806 (1302–2388) |
p | | 0.042 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | 0.001 |
201–300 ng/mL | Apixaban | 24 | 239 (210–274) | 4.9 (3.6–6.1) | 11.7 (8.9–16.6) | 141 (86–178) | 22 (10–38) | 1970 (1509–2608) |
Rivaroxaban | 117 | 240 (221–275) | 7.1 (6.0-8.3) | 18.1 (15.3–21.0) | 102 (70–140) | 9 (6–15) | 1580 (1116–2125) |
p | | 0.548 | < 0.001 | < 0.001 | 0.007 | 0.001 | 0.013 |
> 300 ng/mL | Apixaban | 9 | 329 (313–404) | 5.3 (4.3–6.4) | 12.3 (10.5–14.7) | 106 (83–133) | 15 (8–30) | 1651 (1355–2077) |
Rivaroxaban | 136 | 377 (337–422) | 7.6 (6.1–9.3) | 19.6 (16.2–23.6) | 99 (63–142) | 9 (4–15) | 1533 (956–2106) |
p | | 0.141 | 0.001 | < 0.001 | 0.724 | 0.026 | 0.306 |
Abbreviations: n: number of plasma samples, TTP: time to peak, VI: Velocity index, ETP: endogenous thrombin potential |
Description of outliers
The outliers in peak thrombin were further analysed in order to determine the reason for the high peak thrombin in these patients. There were more outliers in apixaban compared to rivaroxaban treated patients (18 outliers in 16 apixaban vs. 21 outliers in 20 rivaroxaban treated patients, p = 0.003; OR 2.5, 95% CI: 1.3–4.6). Indication for anticoagulation was VTE in 35 patients with concomitant AF in eight. One patient had a prior VTE on anticoagulation with rivaroxaban and another patient had pulmonary embolism three months after cancer surgery. His peak thrombin was within the expected range for the DOAC plasma level another seven months later. Four of the patients with pulmonary embolism (16.7%) had right heart failure which resulted in resuscition in two of them. Only one patient had no history of VTE but AF with history of stroke, resuscitation and coronary artery by-pass. There was a higher incidence of AF (22.9% vs. 3.6%, p < 0.001) and a trend to a higher incidence of pulmonary embolism (60.0% vs. 45.0%, p = 0.082) in VTE patients with at least one outlier compared to VTE patients without outliers. Age, sex and thrombophilia did not influence outliers in patients with VTE (supplementary table 5).
Outliers showed a higher factor VIII activity (240% [IQR 208–287%] vs. 218 % [178–255%], p = 0.006) and a trend to a higher factor IX activity (142% [IQR 124–160%] vs. 134% [IQR 117–151%], p = 0.097) but there was no difference in thrombin time, fibrinogen, factor II, D-dimer and prothrombin fragments.
Bleeding and thrombotic events at follow up
A total of 351 patients who had a follow up visit between April 1st 2020 and May 1st 2021 were analysed. During this time period four thrombotic events occurred. One patient with AF suffered a stroke under anticoagulation with rivaroxaban, one patient had retinal vein thrombosis under rivaroxaban and two patients had DVT after discontinuation of anticoagulation. None of these patients had peak thrombin above the median for their DOAC plasma level. One intracerebral bleeding occurred in a patient under apixaban. Peak thrombin was at the 56th percentile for the DOAC plasma level.
Prediction of normal values
The sensitivity, specificity, positive and negative predictive value and the area under the curve of the ROC for TG parameters within the normal range to predict a DOAC level < 30ng/mL are shown in Table 3 and for the prediction of DOAC plasma levels < 50ng/mL in the supplementary table 6. The resulting ROC curves for the prediction of DOAC levels < 30ng/ml are shown in Fig. 2 and for the prediction of DOAC levels < 50ng/ml in the supplementary Fig. 1.
Table 3
Sensitivity, specificity, positive (PPV) and negative predictive value (NPV) for the prediction of DOAC plasma levels < 30 ng/mL in patients with TG parameters within the normal range.
Assay | DOAC | Sensitivity | PPV | Specificity | NPV | ROC |
| | n | % | n | % | n | % | n | % | AUC |
Lag time | Apixaban | 42/44 | 95.5 | 42/218 | 19.3 | 32/208 | 15.4 | 32/34 | 94.1 | 0.622 |
Rivaroxaban | 199/222 | 89.6 | 199/460 | 43.3 | 219/480 | 45.6 | 219/242 | 90.5 | 0.781 |
TTP | Apixaban | 41/44 | 93.2 | 41/196 | 20.9 | 53/208 | 25.5 | 53/56 | 94.6 | 0.723 |
Rivaroxaban | 190/222 | 85.6 | 190/307 | 61.9 | 363/480 | 75.6 | 363/395 | 91.9 | 0.880 |
Peak | Apixaban | 39/44 | 88.6 | 39/132 | 29.5 | 115/208 | 55.3 | 115/120 | 95.8 | 0.819 |
Rivaroxaban | 196/222 | 88.3 | 196/319 | 61.4 | 357/480 | 74.4 | 357/383 | 93.2 | 0.893 |
Velocity index | Apixaban | 41/44 | 93.2 | 41/164 | 25.0 | 85/208 | 40.9 | 85/88 | 96.6 | 0.789 |
Rivaroxaban | 197/222 | 88.7 | 197/327 | 60.2 | 350/480 | 72.9 | 350/375 | 93.3 | 0.893 |
ETP | Apixaban | 41/44 | 93.2 | 41/179 | 22.9 | 70/208 | 33.7 | 70/73 | 95.9 | 0.739 |
Rivaroxaban | 205/222 | 88.7 | 205/429 | 47.8 | 256/480 | 53.3 | 256/273 | 93.8 | 0.857 |
Abbreviations: AUC: area under the curve, ROC: receiver operator curve |
Thrombin generation in patients with thrombophilia
Thrombin generation in patients with drug levels < 30 ng/ml
Patients with thrombophilia had a shorter lag time (4.3 [IQR 3.7–5.1] vs. 4.8 [4.2–5.8] min, p = 0.001), a shorter TTP (9.3 [IQR 7.4–11.1] min vs. 9.8 [IQR 8.3–11.8] min, p = 0.033) and a trend to a higher ETP (3204 [IQR 2657–3558] nmol/L*min vs. 2999 [IQR 2599–3423] nmol/L*min, p = 0.072). Lag time was shorter in patients with Protein S deficiency (3.4 [2.9–3.7] min vs. 4.8 [IQR 4.2–5.8] min, p = 0.001) and in patients with Factor-V-Leiden-Mutation (4.4 [3.8-5.0] min vs. 4.8 [IQR 4.2–5.8] min, p < 0.005). Patients with Prothrombin-G20210A-Mutation a higher ETP (3456 [IQR 3102–3790] vs. 3013 [IQR 2602–3423], p = 0.002) when only blood samples with a DOAC level < 30 ng/mL were analyzed.
Thrombin generation in patients with drug levels > 30 ng/ml
Apixaban treated patients with thrombophilia had a higher velocity index at plasma levels between 101–200 ng/mL compared to those without thrombophilia (37 [IQR 20–72] nmol/min vs. 21 [IQR 14–41] nmol/min, p = 0.049) and a trend to a shorter lag time at plasma levels between 51–100 ng/mL (5.0 [IQR 3.9–5.7] min vs. 5.6 [4.5–6.4] min, p = 0.058). In addition, patients with a Prothrombin-G20210A-Mutation on apixaban had a shorter time to peak at plasma level between 101–200 ng/mL (7.3 [IQR 6.2–11.3] min vs. 11.3 [9.7–15.0] min, p = 0.038) compared to patients without thrombophilia.
Apart from a shorter lag time at a drug level between 201–300 ng/mL and a Factor-V-Leiden-Mutation (6.6 [IQR 5.3–7.5] vs. 7.3 [6.0-8.5] min, p = 0.052), thrombin generation parameters were not influenced by thrombophilia in rivaroxaban treated patients.
TTP was shorter in apixaban compared to rivaroxaban treated patients with thrombophilia at drug level of 51–100 ng/mL (11.4 [IQR 9.9–12.3] min vs. 13.9 [IQR 11.2–16.8] min, p = 0.007). In addition, lag time and TTP were shorter and peak thrombin and ETP higher in apixaban (n = 20) compared to rivaroxaban (n = 32) treated patients with thrombophilia at DOAC plasma concentrations between 101–200 ng/mL