Rivaroxaban Causes Missed Diagnosis of Protein S Deficiency but Not of Activated Protein C Resistance (Factor V Leiden)

We prospectively reviewed all APCR test results performed by the special coagulation laboratory at Massachusetts General Hospital between March 12, 2014, and September 2, 2016. Results from consecutive patients testing heterozygous for FVL while taking rivaroxaban were recorded in group 1. For each patient in group 1, 3 matching patients were randomly selected among the patients tested on the same day for each of the 3 following control groups (groups 2–4). Group 2 consisted of patients who were also heterozygous for FVL but were not taking rivaroxaban. Group 3 included patients taking rivaroxaban who tested normal (“wild type”) for FVL by DNA testing. Group 4 comprised patients who were not taking rivaroxaban and who had a normal APCR ratio. If control patients tested on the same date as a group 1 patient was not available, control patients were selected from the nearest dates possible that used the same equipment and reagent lot as in group 1, thus reducing bias that could potentially arise from operational differences.

We also identified all patients taking rivaroxaban who underwent protein S testing during the same period. All patients taking rivaroxaban were tested for protein S with both an activity assay and a free-antigen assay.

The patients were evaluated using the standard FVL testing protocol adopted in our institution. It included an aPTT-based APCR assay with dilution in factor V–deficient plasma (Coatest APC Resistance V assay, Chromogenix), on a Star Evolution analyzer, and an FVL DNA assay (Invader assay, Hologic, Boston, Massachusetts). Activated protein C resistance ratios of 2.0 and below were considered abnormal, and to ensure detection of FVL, values of 2.2 or lower underwent further workup by DNA testing. Activated protein C resistance and DNA analysis was performed for all patients taking rivaroxaban.

Protein S functional activity was measured using the STACLOT Protein S assay (Stago) on a Star Evolution analyzer, and free protein S antigen was measured using the Asserachrom Free Protein S assay (Stago). Free protein S antigen levels served as a control to compare to the functional activity of protein S in patients on rivaroxaban, because rivaroxaban does not affect free protein S antigen results. All patients tested for protein S were concomitantly tested for FVL as described above.

Rivaroxaban concentration for each specimen was determined using an anti-Xa assay (Stachrom, Stago). The results were provided in U/mL using a low–molecular weight heparin calibrator (Aniara, West Chester, Ohio), and converted into rivaroxaban concentration using rivaroxaban calibration curves.

The geometric mean APCR ratios of the 4 groups were analyzed using analysis of variance and 2-tailed Student t test assuming equal variances. The mean protein S activity and free protein S antigen were analyzed using 2-tailed Student t test assuming equal variances. P values <.05 were considered statistically significant. Excel software by Microsoft (Redmond, Washington) was used to perform the calculations.

There were 4 groups of 15 patients identified for the APCR analysis (n = 60). The patients included 31 male and 29 female patients between the ages of 20 and 90 years. The geometric means (±SDs) of the APCR ratios in the 4 groups were: group 1 (patients heterozygous for FVL taking rivaroxaban), 1.75 ± 0.12; group 2 (patients heterozygous for FVL not taking rivaroxaban), 1.64 ± 0.09; group 3 (wild-type patients taking rivaroxaban), 2.63 ± 0.23; and group 4 (patients with normal APCR not taking rivaroxaban), 2.49 ± 0.18. Comparing the groups taking and not taking rivaroxaban, the absolute increase in the mean was 0.11 for patients heterozygous for FVL, and 0.14 for the wild-type patients (Figure 1). Individual APCR ratios were all lower than 2.0 for all FVL heterozygous patients, with or without rivaroxaban.

The FVL heterozygous patients taking rivaroxaban had a higher APCR ratio than heterozygous patients not taking rivaroxaban (P = .005). However, these heterozygous patients taking rivaroxaban had a significantly smaller APCR ratio than patients with normal APCR ratios not taking rivaroxaban (P < .001). Although there was a nonsignificant trend for the wild-type patients taking rivaroxaban to have a higher APCR ratio than patients with normal APCR not taking rivaroxaban, a 2-sided t test did not show a statistically significant difference (P = .07). The results are presented in Table 1.

For the protein S study, 32 patients taking rivaroxaban were identified during the abovementioned time period. Of these patients, 8 (25%) were found to concomitantly have FVL, and the remaining 24 (75%) had wild-type factor V. The mean protein S functional activity in patients on rivaroxaban was 124.7% ± 27.9%, and mean free protein S antigen was 83.8% ± 14.8% (Figure 2). The difference between the 2 means was statistically significant (P < .001), presumably because rivaroxaban falsely increases protein S activity but does not affect protein S free antigen. A total of 4 of the 32 patients (12.5%) had low protein S free antigen results (range, 58%–67%). Of these 4 patients, 1 patient (25%) also had FVL, and the remaining 3 patients (75%) had wild-type factor V. Protein S activity was falsely normal in all 4 cases (range, 75%–130%; P < .05). Therefore, unlike APCR, rivaroxaban would have caused a missed diagnosis of all 4 cases of protein S deficiency if using a protein S activity assay alone. The results are presented in Table 2.

Plasma rivaroxaban concentrations of patients in this study ranged from slightly subtherapeutic to high-therapeutic levels.¹⁶  In the APCR study, the rivaroxaban dosage was 20 mg/d in 17 patients (8 FVL heterozygous, 9 wild type), and 15 mg twice daily in 13 patients (7 FVL heterozygous, 6 wild type). In the protein S study, 16 of 32 patients (50%) took 20 mg of rivaroxaban daily (7 of 15 patients with FVL [47%], and 9 of 15 patients with wild type factor V [60%]). Of the remaining 16 patients, 14 (44%) were in their rivaroxaban induction period and took 15 mg twice a day. The information on the daily dosage of the remaining 2 patients (6%) is not available. Both of those patients had wild-type factor V and a therapeutic rivaroxaban concentration.

The results of this study indicate that taking rivaroxaban slightly increases the APCR ratio in patients heterozygous for the FVL mutation. However, the magnitude of this increase did not reach the normal APCR range, and the assay was still able to distinguish the heterozygous patients from normal. No patients were misclassified despite rivaroxaban. The mean increase in the APCR ratio was 0.14 for the wild-type patients, and 0.11 for the heterozygous patients. Figure 1 demonstrates a clear separation between the wild-type and the heterozygous patients taking rivaroxaban. One explanation for this phenomenon could be that diluting the patient's plasma 1:5 was sufficient to reduce the rivaroxaban concentration to a level that caused only a minor interference.

These results are reassuring, and suggest that if specimens are submitted for APCR testing while patients are taking rivaroxaban, the laboratory will provide the correct answer even if the clinician does not inform the laboratory about rivaroxaban, or if DNA testing is not available. The results also suggest that laboratories could proceed with this APCR test despite knowing that rivaroxaban is present. However, as an added precaution, we suggest checking an anti-Xa level to rule out a supratherapeutic rivaroxaban concentration, because levels above 349 ng/mL (anti-Xa 1.51 U/mL by a low–molecular weight heparin curve) were not present in this study. Fortunately, because supratherapeutic specimens were not encountered during the 2.5-year study period, the results suggest that supratherapeutic specimens are not common among patients undergoing hypercoagulability testing.

Rivaroxaban concentrations detected in this study (Table 1) are congruent with the concentrations typically seen in patients taking rivaroxaban. In 1 study, with a 10-mg daily dosage, the steady-state peak was 91 to 196 ng/mL, and the trough was 1 to 38 ng/mL; with a 20-mg daily dosage, the peak was 160 to 360 ng/mL and the trough was 4 to 96 ng/mL (95% confidence interval).¹⁶ 

We did not encounter any patients homozygous for FVL and taking rivaroxaban during the investigation period. However, patients homozygous for FVL exhibit an even greater degree of APCR than heterozygous patients. Therefore, their APCR ratios are even lower than those of heterozygous patients, and fall well below the normal cutoff point.

In a similar study previously conducted by our group, argatroban (a direct thrombin inhibitor) was able to mask the diagnosis of FVL by raising the APCR ratio into the normal range,¹⁷  and dabigatran (direct thrombin inhibitor) substantially raised the APCR ratio.¹⁸  In contrast to these direct thrombin inhibitors, rivaroxaban (an antithrombin-independent factor Xa inhibitor) raised the APCR ratio only slightly, and not into the normal range in any of our cases, highlighting that these different anticoagulants can have different effects on APCR elevation.

Simultaneous testing of functional protein S activity and free protein S antigen in patients on rivaroxaban showed that functional protein S activity was on average 1.5-fold higher than protein S free antigen. This elevation was observed regardless of the presence or absence of FVL, and could be attributed to rivaroxaban. Furthermore, low free protein S antigen occurred in 4 of 32 patients (12.5%) in this study, and the protein S activity assay missed all cases of low protein S during rivaroxaban treatment. Therefore, free protein S antigen is the preferred method in all patients taking rivaroxaban in order to exclude a hereditary protein S deficiency. However, clinicians should keep in mind that in order to detect a rare qualitative type II protein S deficiency (normal levels of protein S free antigen, but decreased protein S function), the protein S activity testing would need to be performed after a patient discontinues rivaroxaban.

The false protein S elevation is likely due to the fact that the protein S activity assay is aPTT based, and protein S prolongs the aPTT by serving as a cofactor for activated protein C–mediated cleavage of factors V and VIII. The degree of aPTT prolongation is proportional to the amount of protein S in the specimen. Rivaroxaban prolongs the aPTT in the assay, causing an overestimation of the amount of protein S present.

Factor V Leiden can cause falsely low protein S activity in some assays. Those assays rely on the patient as the only source of factor V. When the patient has FVL, the patient's factor V resists degradation by the activated protein C/protein S complex, making it seem like there is less protein S present, whereas in reality this apparent decrease in protein S function is caused by the resistance of abnormal factor V to degradation by the activated protein C/protein S complex. The present study avoids such consequences by using a protein S activity assay that supplies exogenous normal factor V, thus minimizing this interference.

The data obtained in this study suggest that the Coatest APCR V assay can be used to test patients anticoagulated with rivaroxaban for the FVL mutation. Because FVL specimens are generally reliably detected at an APCR cutoff of 2.0 or lower, in an effort to ensure 100% sensitivity in detecting FVL, our special coagulation laboratory currently uses a protocol in which all specimens with APCR ratios of 2.2 or lower undergo DNA testing. In the present study, this protocol detected all patients heterozygous for FVL, suggesting that DNA testing of patients with APCRs above 2.2 is unnecessary regardless of rivaroxaban use. As an added precaution, an anti-Xa assay could be performed to confirm that supratherapeutic rivaroxaban levels are not present, because this study included high-therapeutic specimens up to 349 ng/mL rivaroxaban (anti-Xa 1.51 U/mL by a low–molecular weight heparin curve), but supratherapeutic specimens were not encountered (Note added in proof: While the manuscript was in press, we encountered 2 patients heterozygous for factor V Leiden with supratherapeutic rivaroxaban concentrations of 590 ng/mL and 401 ng/mL [anti-Xa 2.48 μ/mL and 1.72 μ/mL by a LMWH curve, respectively]. The APCR assay remained accurate with ratio results of 1.89 and 1.85, respectively. Thus, the APCR assay appears accurate even with supratherapeutic rivaroxaban). Protein S free antigen is preferred instead of protein S activity for patients taking rivaroxaban, with the caveat that protein S activity testing would need to be performed after rivaroxaban has been discontinued in order to detect a rare qualitative type II protein S deficiency.