Effects of Indirect- and Direct-Acting Anticoagulants on Lupus Anticoagulant Assays: A Large, Retrospective Study at a Coagulation Reference Laboratory

Abstract

Materials and Methods

Due to the lack of specificity of clinical features in the diagnosis of the antiphospholipid syndrome (APLS), laboratory investigations for antiphospholipid antibodies usually play a critical role.¹ Laboratory-based criteria include functional coagulation assays for the detection of a lupus anticoagulant (LAC) and solid-phase assays to detect immunoglobulin G (IgG)/immunoglobulin M (IgM) antibodies to cardiolipin (ACL) and β2 glycoprotein I (anti-β2GPI).² LAC is an antiphospholipid antibody, which prolongs the in vitro coagulation time and is not reversed by mixing with normal plasma but is reversed when extra phospholipid is added.^3,4 Numerous organizations, including the International Society of Haemostasis and Thrombosis (ISTH) and the Clinical and Laboratory Standards Institute (CLSI), recommend that two different assays with distinct performance principles be undertaken to detect LACs, due to antibody heterogeneity and reagent variations.^5-8

The introduction of direct oral anticoagulants (DOACs), including direct thrombin inhibitors (DTIs) and direct factor Xa inhibitors (DFXaIs), as alternatives to warfarin in patients with atrial fibrillation and venous thromboembolic events has complicated the interpretation of traditional coagulation assays, including panels used for the diagnosis of LAC.^9-12 Bonar et al¹³ investigated the interference of LAC assays by dabigatran by sending in vivo and ex vivo samples to 72 laboratories participating in the Royal College of Pathologists of Australasia Quality Assurance Programs hematology program. The study found that the activated partial thromboplastin time (APTT) assay (Dade Actin FS, Siemens, Marburg, Germany) showed some sensitivity to dabigatran. Interestingly, the authors reported a near-linear relationship between the dabigatran concentration in the six survey samples and the dilute Russell viper venom time (dRVVT) reported by the laboratories; there was also a proportional increase in the dRVVT screen and confirm results (Staclot DRVV Screen and Confirm; Diagnostica Stago, Asnières-sur-Seine, France) with increasing dabigatran concentrations and a corresponding increase in the dRVVT screen/confirm ratio. The authors, however, cautioned that the relatively low number of respondents and significant scatter seen in the results limit the usefulness of the dRVVT in quantification of dabigatran levels.¹³

Martinuzzo et al¹⁴ reported that dabigatran caused prolongation of the APTT (APTT SP, Instrumentation Laboratory, Bedford, MA) and dRVVT (LAC Screen and LAC Confirm, Instrumentation Laboratory), a positive screen/confirm normalized ratio (NR), and prolongation of the silica clotting time (SCT) Screen and Confirm (Instrumentation Laboratory). An Austrian multicenter trial confirmed that dabigatran causes interference with the prothrombin time (PT), APTT, and the dRVVT screen, confirm, and NR tests.¹⁵ The authors also stated that normalization of dRVVT data failed to correct the false-positive LAC test results at high dabigatran concentrations. Finally, Kim et al¹⁶ have reported on their findings of falsely positive hexagonal phase lipid neutralization results caused by dabigatran using the Staclot LA assay (Diagnostica Stago); the magnitude of the effect was dependent on the concentration of dabigatran in the sample.

Studies have also shown interference by rivaroxaban on LAC testing. Góralczyk et al¹⁷ reported false-positive screening APTT (PTT-LA, Diagnostica Stago) and dRVVT (HemosIL dRVVT Screen/HemosIL dRVVT Confirm, Instrumentation Laboratory) assays in 10 patients treated with the factor Xa inhibitor; a 24-hour washout was required for the tests to become negative. Mani et al¹⁸ found that rivaroxaban caused a significant increase in dRVVT (HemosIL dRVVT Screen/HemosIL dRVVT Confirm) 2 hours after low-dose rivaroxaban administration compared with the data prior to administration or 12 hours after administration, whereas there was no difference seen with the MIXCON-LA assay (HemosIL APTT-SP and SynthAFax, Instrumentation Laboratory). Martinuzzo et al¹⁴ also found that rivaroxaban led to positive dRVVT screen, mixing, and confirmatory tests (HemosIL dRVVT Screen/HemosIL dRVVT Confirm); prolonged APTT (APTT SP) and SCT screens; and a negative screen/confirm NR (SCT Screen and Confirm). Of the 21 patients in the EINSTEIN study who had LAC testing on rivaroxaban, 19 had a positive screening test, at times ranging between 1 and 19.5 hours after the rivaroxaban dose.¹⁹ The dRVVT was the most susceptible to interference, showing no or only partial correction with 1:1 mixing in all but one patient. The dilute thromboplastin time was also affected. Five patients had a prolonged screening APTT (Triniclot, TCoag, Diagnostica Stago), two patients had a positive Kaolin clotting time, and none had a positive APTT (Staclot APTT, Diagnostica Stago) test. Douxfils et al²⁰ have also reported on the effects of apixaban on LAC assays, describing a concentration-dependent prolongation of the PTT-LA, Staclot LA (Diagnostica Stago), dRVVT Screen and Confirm, and dilute PT (Triniclot PT Excel, TCoag, Diagnostica Stago).

Results

The effects of vitamin K antagonists, such as warfarin, and heparin on LAC assays have been documented extensively. Briefly, warfarin may cause false-positive LAC screening results, especially with the dilute PTs and dRVVT and, to a lesser extent, with the APTT due to reduced coagulation factor levels (vitamin K–dependent factors II, VII, IX, and X); a 50:50 mixing study usually but not always corrects the clotting time, and false-positive mixing studies can also occur with warfarin.^21,22 Unfractionated heparin may also cause a false-positive APTT with a failure to correct on mixing studies. However, the presence of heparin can be detected by use of a thrombin time (TT) and reptilase time; if heparin is present in the sample, it can be neutralized or degraded by agents, such as polybrene or heparinase. Certain assay kits already contain heparin neutralizers, which detect LAC in the presence of heparin.^21,22

Most studies, however, have shown no interference by anticoagulants on solid-phase assays or enzyme-linked immunosorbent assays (ELISAs) for ACL or anti-β2GPI.^14,17,19 This study aims to determine the rates of LAC test positivity based on anticoagulant type to assist with interpretation of LAC panels in patients being treated with traditional and direct oral anticoagulants at a large coagulation reference laboratory.

A retrospective database review was performed to identify all LAC and antiphospholipid antibody panels performed at the Institute for Transfusion Medicine from November 23, 2012, to November 22, 2015. The hypercoagulable workup included testing for antiphospholipid antibodies by ELISA, including anti-β2GPI IgG and IgM antibodies, ACL IgG and IgM antibodies, and testing for LAC with three tests using different methods: (1) LAC-sensitive dilute APTT (dAPTT) with a hexagonal phase lipid neutralization (hexagonal PL) test, (2) dRVVT screen and confirm ratio, and (3) the tissue thromboplastin inhibition index (TTI) with a PT.²³ The TTI assay result was disregarded if the PT was above the upper limit of the reference range, since the TTI assay is based on 1:50 and 1:500 dilutions of the PT reagents and there was no validated mixing study for the TTI assay.

❚Figure 1❚

Open in new tabDownload slide

Positivity rates for lupus anticoagulant (LAC) assays stratified by anticoagulation treatment type. APIX, apixaban; ARG, argatroban; DAB, dabigatran; dAPTT, dilute activated partial thromboplastin time; dRVVT, dilute Russell viper venom time; FONDA, fondaparinux; LMWH, low molecular weight heparin; NO TX, no treatment; PL, phospholipid; PT, prothrombin time; RIV, rivaroxaban; TTI, tissue thromboplastin inhibition index; UFH, unfractionated heparin; WARF, warfarin. *P < .05 vs NO TX group for all panels. In total, 3,843 TTI results were excluded due to a prolonged PT (>11.5 seconds). The overall LAC panel result was based on Clinical and Laboratory Standards Institute guidelines.

Positivity rates for lupus anticoagulant (LAC) assays stratified by anticoagulation treatment type. APIX, apixaban; ARG, argatroban; DAB, dabigatran; dAPTT, dilute activated partial thromboplastin time; dRVVT, dilute Russell viper venom time; FONDA, fondaparinux; LMWH, low molecular weight heparin; NO TX, no treatment; PL, phospholipid; PT, prothrombin time; RIV, rivaroxaban; TTI, tissue thromboplastin inhibition index; UFH, unfractionated heparin; WARF, warfarin. *P < .05 vs NO TX group for all panels. In total, 3,843 TTI results were excluded due to a prolonged PT (>11.5 seconds). The overall LAC panel result was based on Clinical and Laboratory Standards Institute guidelines.

Discussion

In total, 7,721 profiles, which contained at least one of the three LAC tests, from 6,334 patients were extracted from the database. There were 1,587 repeat panels performed on 1,144 patients. The mean ± SD patient age was 53.9 ± 18.3 years (range, 1-96 years). Fifty-five percent of the panels were from female patients. A significant number of TTI results (3,843) were not interpretable because the PT was prolonged: 31% for NO TX, 98% for WARF, 54% for UFH, 19% for LMWH, 26% for FONDA, 100% for ARG, 69% for DAB, 39% for RIV, and 33% for APIX. Based on CLSI guidelines and local reference ranges, more than half of all LAC panels performed on patients treated concurrently with DTIs or DFXaIs were positive (83% for ARG, 58% for DAB, 72% for RIV, and 53% for APIX vs 29% in the no-treatment group, P < .05). ❚Figure 1❚ highlights the frequency of positive results by assay type for each treatment group. ❚ Table 2❚ and ❚Figure 2❚ show the breakdown of positivity rates for the three laboratory criteria for a diagnosis of APLS. There was no significant difference in positivity rates when repeat panels were excluded.

Compared with patients not on documented anticoagulation, DTIs caused a higher positivity rate for the dAPTT + hexagonal PL combined test (63% vs 8%, P < .001) and the TTI + PT combined test (75% vs 11%, P < .001) with no significant interference in the dRVV ratio (26% vs 20%, P = .380). However, the TTI assay could only be interpreted in four panels, with the remainder being associated with a prolonged PT. There was significant heterogeneity in assay positivity rates when both DTI agents were compared; DAB had a lower positivity rate for the dAPTT + hexagonal PL combined test (33% vs 72% for ARG, P = .016), and there was a trend toward a lower overall positivity rate for DAB, but this was not significantly different (58% for DAB vs 83% for ARG, P = .083). There was no significant difference in ACL or β2GPI positivity rates between the DTI groups and the no-treatment group.

DFXaIs affected mainly the TTI + PT combined test (37% vs 11% positive for NO TX, P < .001) and the dRVV ratio (64% vs 20% positive for NO TX, P < .001) with less observed interference in the dAPTT + hexagonal PL combined test (14% vs 8% positive for NO TX, P < .001). There were also significant differences between RIV and APIX: RIV had a higher positivity rate overall (72% vs 53%, P = .034) and a higher positivity rate for the dRVV ratio (66% vs 27%, P < .001). The differences between both agents for the TTI + PT combined test (55% vs 36%, P = .092) and dAPTT + hexagonal PL combined test (14% vs 17%, P = .708) did not reach statistical significance. Although not statistically significant, there was a higher β2GPI positivity rate in patients treated with RIV (9.7% vs 5.5% for NO TX, P = .059).

Sixty patients, who had LAC testing while on a DOAC (RIV, APIX, or DAB), also had repeat testing after the DOAC was discontinued (mean ± SD interval between tests was 129 ± 107 days; range, 1-547 days); two patients were on ARG, one on DAB, 56 on RIV, and one on APIX. The initial LAC panel was positive in 50 (83%) of these patients; when the panel was repeated after the DOAC was discontinued, only 13 (26%) of these patients had a positive LAC result. This rate was similar to the overall rate of LAC positivity in the no-treatment, UFH, LMWH, and FONDA groups. Thirty-eight patients had a negative LAC panel while not on a DOAC agent (23 patients were not on treatment, six were on WARF, eight were on UFH, and one was on FONDA) and also had repeat testing after a DOAC was commenced. On repeat testing after the DOAC agent was commenced, 28 (74%) had a positive LAC result. The mean ± SD interval between tests was 177 ± 163 days (range, 35-820 days).

Warfarin therapy was associated with a higher positivity rate overall (56% vs 29%, P < .001), as well as for the dAPTT + hexagonal PL combined test (30% vs 8%, P < .001) and the dRVV ratio (41% vs 20%, P < .001) compared to the no-treatment group. Only 30 TTI assays could be interpreted since the remainder of panels demonstrated a prolonged PT. There was no difference seen in positivity rates for the TTI + PT combined test. Panels performed on patients treated with warfarin also displayed a higher ACL positivity rate (13.8% vs 10.8% in the no-treatment group, P = .023).

Comparing the heparin agents to the no-treatment group, UFH was associated with a higher rate of positive results overall for the LAC panel (36% vs 29%, P < .001) and, in particular, for the dAPTT + hexagonal PL combined test (12% vs 8%, P < .001) and TTI + PT combined test (34% vs 11%, P < .001); positivity rates were not significantly different between LMWH or FONDA and the no-treatment group except for a higher dRVV ratio positivity rate in patients on FONDA (28% vs 20%, P = .044). There was a positive correlation between the TT result and the rate of positivity for the hexagonal PL (pseudo-R² = 0.01, P < .001) and TTI (pseudo-R² = 0.06, P < .001) assays. UFH was associated with a lower ACL positivity rate (7.4% vs 10.8% in the no-treatment group, P = .006), while FONDA was associated with a twofold higher ACL positivity rate (21.7% vs 10.8% in the no-treatment group, P = .009).

There were 782 dRVVT mixing studies performed (including 11% in the NO TX group, 70% in the WARF group, 11% in the UFH group, and the remainder in patients on DOACs). Almost all (97%) dRVV mixing studies were performed when the PT was prolonged. There were also 2,881 dAPTT mixing studies performed due to a suspicion of factor deficiency causing a positive hexagonal PL result; most were performed in patients on WARF (55%), NO TX (24%), UFH (12%), and RIV (5%), with the remainder divided evenly among the other drug groups. If the dAPTT mix result was included in interpretation of the dAPTT + hexagonal PL combined test, the positivity rate fell significantly: 2.4% in NO TX, 4.7% in WARF, 3.3% in UFH, 1.9% in LMWH, 3.2% in FONDA, 50.0% in ARG, 30.8% in DAB, 9.4% in RIV, and 6.7% in APIX.

This large, retrospective study investigated the interference effects of indirect and direct-acting anticoagulant agents on assays used to detect the presence of LAC. Significantly higher positivity rates (>50%) were observed in patients treated with DFXaIs, DTIs, and warfarin compared with patients not on documented anticoagulation (29%) or patients on heparin (30%-36%).

There was evidence of interference by RIV in all three tests in the LAC panel, based on the higher positivity rates vs the no-treatment group. APIX significantly affected the TTI + PT test to a greater degree than RIV, and there was a nonsignificant trend toward higher positivity rates in both the dAPTT + hexagonal PT test and dRVV ratio in patients on APIX compared with the no-treatment group. The effect of DFXaIs, particularly rivaroxaban, on the dRVV ratio has been studied extensively in the literature and was supported by the results of this study.^14,25,26 This interference effect is expected since Russell viper venom activates factor X to factor Xa, which is subsequently inhibited by DFXaIs. However, it is noteworthy that the effect of DFXaIs on the dRVVT is drug dependent and not necessarily a class effect, since APIX did not significantly increase the positivity rate compared with the no-treatment group. This response heterogeneity of coagulation assays to different DFXaIs has been documented previously.²⁷ Prior studies have also reported on the interference of DFXaIs with PT-based assays, but this effect appears to be reagent dependent, since the interference effect of DFXaIs on the PT reagent (Innovin, Siemens) at our laboratory was less pronounced compared with the effect of warfarin or DTIs.²⁷ Two-thirds of the TTI results would have been included in the overall LAC panel interpretation for patients on DFXaIs since the PT results were within the reference range, leading to a false-positive interpretation in a significant proportion (half) of these cases. It is likely that the interference effect of DFXaIs becomes more pronounced as the PT reagent is diluted.

❚Figure 2❚

Open in new tabDownload slide

Positivity rates for select combinations of the three laboratory criteria for antiphospholipid syndrome stratified by anticoagulation treatment type. Only 4,268 panels had testing performed for both lupus anticoagulant and ACL/β2GPI antibodies. ACL, anticardiolipin antibodies (immunoglobulin G [IgG] or immunoglobulin M [IgM]); APIX, apixaban; ARG, argatroban; β2GPI, anti–β2 glycoprotein I antibodies (IgG or IgM); DAB, dabigatran; FONDA, fondaparinux; LMWH, low molecular weight heparin; NO TX, no treatment; RIV, rivaroxaban; UFH, unfractionated heparin; WARF, warfarin. *P < .05 vs NO TX group for all panels.

Open in new tabDownload slide

The TTI + PT combined test was not useful in patients on DTIs since most of panels (92%) performed with a DTI on board demonstrated a prolonged PT. Both ARG and DAB caused significant interference in the dAPTT + hexagonal PL combined test but to varying extents (ARG was associated with a greater than twofold higher positivity rate compared with DAB). The overall LAC panel positivity rate was also 1.5-fold higher for ARG compared with DAB, again highlighting that the interference effect varies among agents within the same class. Too few panels in patients on DTIs included testing for ACL and β2GPI to draw any meaningful conclusions.

WARF was also associated with a twofold higher LAC positivity rate compared with the no-treatment group. Most (98%) of the TTI assays performed in patients on WARF were not interpretable due to a prolonged PT, hence limiting the usefulness of this assay in this group of patients. Patients treated with warfarin had a twofold higher positivity rate for the dRVV ratio vs the no-treatment group. Warfarin interference in the dRVV ratio due to low factor X levels has been reported in the literature, and this effect is corroborated by the findings in this study.²⁸ Prior reports have noted that APTT-based phospholipid neutralization assays such as Staclot LA tend not to show spurious positive results in plasmas with vitamin K antagonists.²⁹ This finding is due to the use of a 50/50 mix with normal human plasma in the Staclot LA assay, which theoretically negates the factor deficiency issue caused by warfarin. However, the findings of this retrospective study contradict these claims; there was a more than threefold higher positivity rate observed in the dAPTT + hexagonal PL combined test compared with the no-treatment group. Only when an external dAPTT mixing study was performed did the positivity rate fall to a level close to the no-treatment group (4.7% vs 2.4%, P < .001). It is possible that the ratio of normal plasma to patient plasma in the Staclot LA test system is insufficient to completely counteract the effect of vitamin K antagonists. There was a 30% higher ACL positivity rate in the WARF group compared with the no-treatment group, and a higher proportion was also positive for both LAC and an ACL/β2GPI antibody. This is likely because most, if not all, patients on vitamin K antagonists have a documented history of thrombosis, with a corresponding higher prevalence of APLS.

References

FONDA was associated with a marginally higher dRVV ratio positivity rate and a twofold higher positivity rate for the ACL assay. This was likely due to the fact that most of these patients on FONDA had a suspected diagnosis of heparin-induced thrombocytopenia (HIT), and there may have been nonspecific antiphospholipid antibodies or cross-reacting antibodies present as part of the HIT syndrome. Although Martinuzzo et al¹⁴ previously reported that patients receiving enoxaparin had a high prevalence of APTT (APTT SP) and DRVVT (DRVVT1 Screen and Confirm, LAC Screen and Confirm) false-positive results, our study failed to show a significantly higher positivity rate for LMWH compared with patients on no treatment for all three LAC tests. Hence, if testing for LAC were indicated for a patient on a DOAC agent, who is at risk for recurrent thrombosis off anticoagulation, transitioning the patient to a LMWH briefly to perform LAC testing would be considered a reasonable, evidence-based approach. A recent study suggests that a 24-hour washout period is needed to negate the effects of DOACs on coagulation assays.¹⁷

For patients with repeat testing data available, approximately 75% with an initial positive LAC panel while on a DOAC agent reverted to a negative result on subsequent testing when the DOAC was held or discontinued. The 26% positivity rate after the DOAC agent was discontinued or held was similar to the rate seen in the no-treatment and LMWH groups. Likewise, the same proportion of patients who were initially negative, while not on a DOAC, had subsequent positive LAC results when testing was repeated on a DOAC. These results further point toward assay interference as opposed to the presence of a true LAC.

In accordance with CLSI guidelines, mixing studies were only performed when there was a suspicion of a factor deficiency.²⁴ This is in contrast to ISTH guidelines, which recommend performing mixing studies immediately after a prolonged screening test and prior to a confirmatory test.⁵ The rationale for omitting the mixing studies unless a there is a clear-cut indication is based on the possibility of masking a weak antiphospholipid antibody due to a dilutional effect. Nonetheless, this data set showed that routine performance of the dAPTT mix assay lowers the positivity rate of the dAPTT + hexagonal PL test, which may correspond to an improved specificity of the test. Although the hexagonal PL assay uses normal plasma to correct factor deficiencies, a significantly higher positivity rate for this test was observed in the WARF group. This may either have represented assay interference or true positivity since these patients were likely to have a history of thromboembolism necessitating anticoagulant therapy. However, the indication for RIV and APIX in this population was the same, and yet the positivity rate was half of that seen in the WARF group, thus making the former explanation more plausible.

The TTI + PT test was generally not helpful in patients on WARF and DTIs since most of these patients had a prolonged PT. Use of a validated PT/TTI mixing study in these cases may improve the applicability of this test to a general patient population. The TTI result also may have been misleading in patients on DFXaIs since many did not have a prolonged PT, yet the positivity rate for the TTI + PT test was three to five times higher than the rate seen in the no-treatment group. The inhibitory effect of the DFXaI on the assay was more pronounced when the PT reagent was diluted.

This study does have limitations based on the retrospective nature of data collection and the fact that the database is not complete since it relies on submission of appropriate medication information on the test requisition form or follow-up with ordering physicians, when possible. There was a markedly higher LAC positivity rate in the no-treatment group compared with estimates from population studies.³⁰ It is unlikely that all anticoagulant medications have been captured in the database, and drug interference effects may potentially explain a subset of positive LAC panels in the no-treatment group. Patients in the no-treatment group may also have been on antiplatelet agents for arterial thombotic events, which are also associated with APLS. Blood concentrations of DTIs and DFXaIs were not measured routinely, and data on the time of drug administration were unavailable, so correlations between drug levels and positivity rates could not be done. Finally, patients with LAC are more likely to have repeated testing for diagnostic purposes and to guide therapeutic decision making; it is therefore likely that these patients were overrepresented when all panels were analyzed. However, there was no significant difference in positivity rates overall or for the individual assays when repeat panels were excluded from the analysis.

In conclusion, we recommend that LAC testing should not be performed on patients treated with DOACs, and results of tests performed while on other anticoagulant therapies, except LMWH and FONDA, should be interpreted with caution. If testing is performed while on a DOAC, a disclaimer should be included with the interpretation stating that these agents may interfere with coagulation assays routinely used to test for a LAC, making the interpretation invalid and that repeat testing after at least a 24-hour washout period would be recommended.¹⁷ Since anticoagulant information is not always readily available, especially for samples sent to reference laboratories, standard disclaimers may be warranted for all positive LAC results advising about the effects of DOACs. The College of American Pathologists already requires the warning of clinicians about the interfering effects of anticoagulants on APTT-Mix results.³¹ Alternately, testing with assays or reagents that have demonstrated no known interference by DOAC agents, such as the Taipan venom time/Ecarin clotting time ratio or the Textarin time, may be useful, but these are not commercially available at present.²⁶ Testing for ACL and β2GPI antibodies may be considered in most patients with suspected APLS. However, the possibility of cross-reactivity of antibodies in patients with HIT should be taken into account when interpreting results, especially in patients treated with FONDA and ARG. It is currently cost-prohibitive to screen all samples for the presence of therapeutic levels of a DOAC, but this may become possible in the future.

Author notes