Ann Hematol DOI 10.1007/s00277-015-2374-3

ORIGINAL ARTICLE

Monitoring anticoagulant therapy with vitamin K antagonists in patients with antiphospholipid syndrome Mecki Isert 1 & Wolfgang Miesbach 2 & Gundolf Schüttfort 1 & Yvonne Weil 1 & Vanessa Tirneci 1 & Alexander Kasper 1 & Adele Weber 1 & Edelgard Lindhoff-Last 1 & Eva Herrmann 3 & Birgit Linnemann 1

Received: 28 December 2014 / Accepted: 30 March 2015 # Springer-Verlag Berlin Heidelberg 2015

Abstract Because of the possible interference of antiphospholipid antibodies (APL) with the phospholipid component of thromboplastin reagents, concerns have been raised about the validity of international normalized ratio (INR) testing to monitor anticoagulant therapy with vitamin K antagonists in patients with antiphospholipid syndrome (APS). To investigate the reliability of the INR, we determined the INR using various prothrombin time (PT) assays and compared the results with those of a chromogenic factor X (CFX) assay. The study cohort consisted of 40 APS patients and 100 APL-negative patients who were on anticoagulant therapy for reasons other than APS. The agreement (i.e. the percentage of patients with a difference ≤0.5 INR units) between the PTderived INR and CFX-derived INR equivalents was only moderate in both patient groups. The best agreement with CFX-derived INR equivalents was observed for the Thromborel S reagent in APS patients (69.1 %) and for Neoplastin Plus in APL-negative patients (72.0 %). Regarding the results for the point-of-care system CoaguChek XS, an agreement between the INR and the CFX-derived INR equivalent was less frequently observed in the APS patients (55.6 vs. 67.8 %; p=0.050). When considering all 3058 pairs of INR tests within the international sensitivity index (ISI)-

* Birgit Linnemann [email protected] 1

Division of Vascular Medicine, Department of Internal Medicine, Goethe University Hospital, Theodor-Stern-Kai 7, 60590 Frankfurt/ Main, Germany

2

Frankfurt Haemophilia Center, Institute of Transfusion Medicine, Goethe University Hospital, Frankfurt/Main, Germany

3

Institute of Biostatistics and Math, Modelling, Faculty of Medicine, Goethe University Hospital, Frankfurt/Main, Germany

calibrated range of 1.5 to 4.5 s, we did not observe a higher variability of INR values in either the APS patient group or the subgroup of APS patients positive for lupus coagulants compared with the APL-negative controls. In conclusion, monitoring vitamin K antagonists (VKA) therapy with laboratory INR measurements seems to be suitable for the majority of APS patients. Keywords Antiphospholipid syndrome . Lupus anticoagulants . International normalized ratio . Chromogenic factor X assay . Anticoagulant therapy

Introduction Antiphospholipid syndrome (APS) is an autoimmune disorder characterized by the occurrence of vascular thrombosis or pregnancy complications [1, 2]. Anticoagulant therapy has been shown to be highly effective in preventing recurrent thrombotic events. Despite the advantages of anticoagulant therapy, APS patients remain at high risk of developing future thromboembolic events. Pengo et al. followed 160 patients with APS who were triple-positive (i.e. the presence of lupus anticoagulant, anticardiolipin and anti-beta-2-glycoprotein-I antibodies) and reported a cumulative incidence of 29 % for subsequent thromboembolism among APS patients during long-term oral anticoagulant treatment [3]. Vitamin K antagonists (VKA) are the therapy of choice for a long-term anticoagulant therapy, but adequate dosing requires rigid laboratory monitoring. VKA therapy is monitored using the international normalized ratio (INR), which is derived from the prothrombin time (PT) and the appropriate international sensitivity index (ISI) of the thromboplastin used in the PT test system [4, 5]. However, some PT assays have been demonstrated to be sensitive to the effects of lupus

Ann Hematol

anticoagulants (LA), leading to a prolonged PT and consequently false INR values [6, 7]. Tissue factor is the active ingredient in thromboplastin reagents and is used in PT assays to trigger coagulation. Thromboplastin reagents contain either recombinant tissue factor or tissue factor extracted from the brain and placental sources [8]. Thromboplastins also vary in the type and amount of the phospholipid component, which has been shown to influence the results of PT testing [9]. With regard to the phospholipid composition used in the test system, the presence of antiphospholipid antibodies (APL) in a patient’s plasma sample may have important implications for the dosing of VKA therapy due to an overestimation of the INR. Chromogenic factor X (CFX) assays are phospholipid-independent. Measuring factor Xa activity with the CFX assay may therefore provide benefits to APS patients because the test results are not influenced by the presence of APL. To date, few studies have compared INR values and factor X activities in patients with and without APS [10–14]. The patient number in some studies was only small [12, 13], and the results are conflicting. Moreover, because CFX assays were until recently costly and not widespread in use, the therapeutic range of anticoagulant therapy has not been properly evaluated [15]. The aim of the present study was to investigate the reliability of the INR for monitoring anticoagulant therapy in patients with definitive APS. For this purpose, we measured the INR using different laboratory assays and a point-of-care system and compared the results of the INR testing with those of a phospholipid-independent CFX assay.

department. We enrolled 20 patients in each predefined INR category: INR 1.50–1.99, 2.00–2.49, 2.50–2.99, 3.00–3.49 and ≥3.50, according to the INR obtained by the use of our routine thromboplastin reagent Neoplastin Plus (Roche Diagnostics, Basel, Switzerland). The stratification of the INR was performed to ensure a reliable regression analysis between the INR and CFX levels over a wide range of INR values. Blood sampling in the APS patients was performed upon enrolment in the study and after 1, 2, 3, 6 and 12 months of follow-up. Of the 40 APS patients initially enrolled, two were lost to followup, one female became pregnant and another patient suffered a cerebral infarction and died due to intracerebral haemorrhage. Thus, a total of 224 paired INR and CFX measurements from the APS patients were finally analysed. All of the patients were advised to take their VKA dose in the afternoon or evening, such that the last dose was taken at least 10 to 12 h before blood sampling. Blood sampling Venous blood samples were collected with the S-Monovette system (Sarstedt, Nümbrecht, Germany) containing 0.106 mol/L of sodium citrate solution. Blood was centrifuged as soon as possible but no later than 30 min after collection for 2×15 min at 2500g (Rotina 420 R, Hettich Lab Technology, Tuttlingen, Germany). INR testing was performed immediately using our routine assay. The remaining supernatant plasma was stored in aliquots in capped plastic tubes, which were coded and frozen at −70 °C until further analysis. Prothrombin time and international normalized ratio

Methods Participants The study cohort consisted of 40 patients diagnosed with APS and treated with VKA because of a documented history of either venous or arterial thrombosis. A total of 100 patients treated with VKA for reasons other than APS (e.g. secondary prevention of VTE, atrial fibrillation, heart valve prosthesis) served as the control group. All procedures followed were in accordance with the ethical standards of the responsible committee on human experimentation (institutional and national) and with the Helsinki Declaration of 1975, as revised in 2008. All patients received written and oral study information and gave written informed consent to participate. Our institutional ethics committee approved the study, and the study was registered at http://ClinicalTrials.gov (NCT01660061). All study participants provided blood samples for competitive INR and CFX measurements. Control subjects were eligible for this study depending on the INR value determined when the patients attended their scheduled appointment of routine INR monitoring at our university hospital’s outpatient

INR measurements were performed using four commercial PT assays and thromboplastin reagents that were most widely used in Europe and North America. Aside from Neoplastin Plus (Lot 641838, international sensitivity index (ISI) 1.32), which was extracted from a rabbit brain, one reagent was extracted from a human placenta (Thromborel S, Lot 545362, ISI 1.03, Siemens Healthcare, Marburg, Germany) and two were based on recombinant human tissue factor (RecombiPlasTin, Lot N1104147, ISI 1.01, Instrumentation Laboratory, Bedford, USA; Innovin, Lot 539157, ISI 0.89, Siemens Healthcare). Because of the instrument-specific ISI values, calibration for each thromboplastin was performed according to the recommendations of the manufacturer. INR measurements were performed on the STA-R Evolution® coagulation analyser (Diagnostica Stago, Asnières sur Seine, France) using the Neoplastin Plus reagent, on the ACL Top® 700 (Instrumentation Laboratory) using RecombiPlasTin and on the Behring Coagulation System® BCS (Siemens Healthcare) using Thromborel S and Innovin. The PT values obtained were converted into an INR value by dividing a patient’s PT by the mean PT of control samples and by

Ann Hematol

correcting this PT ratio by the manufacturer’s instrumentspecific ISI. To exclude substantial differences in the INR tested in fresh and frozen plasma samples, we determined the INR in fresh and thawed plasma samples. The frozen plasma samples were thawed at 37 °C for approximately 15 min, gently mixed and tested within 1 h of thawing. The INR values of fresh and frozen plasma samples were nearly identical (Pearson’s correlation coefficient r =0.980; coefficient of variation CV = 2.8 %) when the two measurements were performed within a 3-month period. For the subsequent analyses, frozen plasma samples were used for INR measurements with the different laboratory PT assays and for CFX measurements. To measure the INR with the CoaguChek® XS system (Roche Diagnostics), venous blood was dropped on a test strip (LOT 367, 416, 437; ISI=1.00). Plesch et al. had demonstrated before that there was no significant deviation between results from capillary with those from corresponding venous blood samples on the CoaguChek XS system [16]. This system has an integrated quality control function. Additionally, we performed external quality controls using CoaguChek® XS Pro PT Control liquids (Roche Diagnostics) at least once weekly. Chromogenic factor X assay CFX was measured on the ACLTop® 700 coagulometer using the Biophen Factor X reagent kit (Hyphen BioMed; Neuvillesur-Oise, France). In a two-step reaction, factor X is activated to factor Xa in the presence of calcium using Russell’s viper venom as an activator. Subsequently, factor Xa hydrolyses the chromogenic substrate SXa-11, thus liberating the chromophoric group p-nitroaniline. The intensity of the colour is read photometrically at 405 nm and is proportional to the factor X activity of the sample. A standard curve is created using serial dilutions of reference plasma containing a known amount of factor X. The precision of this method was determined by means of the coefficient of variation (CV) between two measurements and revealed a CV of 8.8 %. Antiphospholipid antibodies All of the patients were tested for the presence of APL to ensure that none of the control subjects were APL positive and to determine the type of APL in the APS patient group. APL testing was performed according to the international criteria of the ISTH (International Society on Thrombosis and Haemostasis) Scientific and Standardisation Committee [17, 18]. The presence of LA was diagnosed by combining different tests [dilute Russell viper venom time (dRVVT) (LAC Screen and LAC Confirm; Instrumentation Laboratory), a lupus-sensitive aPTT (Haemoliance SynthAFax aPTT Reagent; Instrumentation Laboratory) and the Mixcon-LA for

mixing and confirmation procedures (Instrumentation Laboratory)]. A fully automated chemiluminescent immunoassay was used on the ACL AcuStar® to measure antibodies of the IgG and IgM isotypes against cardiolipin (ACL) (HemosIL Acustar Anti-Cardiolipin IgG and IgM; Instrumentation Laboratory) and beta-2-glycoprotein-I (β2GPI) (HemosIL Acustar Anti-β2 Glycoprotein-I IgG and IgM, Instrumentation Laboratory). Only antibody titres above 20 U/mL were considered positive according to the current diagnostic criteria of APS [19]. Statistical analyses Statistical analyses were performed using IBM SPSS Statistics version 21 (Armonk, USA). In addition to calculating descriptive statistics (including frequencies, mean, standard deviation, median and range), we performed a chi-squared test in cross-tabulations and the Mann–Whitney U test for the comparison of metric variables. The criterion for statistical significance was a p value less than 0.05. A linear regression model was used to determine the agreement between paired INR values from the fresh and frozen plasma samples. Pearson’s correlation coefficient (r) and the CV were calculated from pairs of test results. Various nonlinear regression models were calculated to find the best-fit relationship between the test results of the CFX and INR assays. The goodness of fit was expressed as the coefficient of determination R2. According to the best-fit regression curve, INR equivalents were calculated for each CFX value, and the results compared with the INR values obtained from the different PT assays. Bland–Altman plots were drawn to illustrate the agreement between the INR and CFX-derived INR equivalents. The mean difference and standard deviation of the differences are also indicated in the plot. We defined a deviation greater than 0.5 INR units between two methods of INR determination as the limit of acceptance and calculated the percentage of patients in agreement of INR test results (i.e. INR difference ≤0.5) between two test systems. Because the variability of INR testing increases with the absolute value of INR, we restricted our final analysis of paired INRs to values within the ISI-calibrated range (i.e. 1.5–4.5).

Results The APS cohort consisted of 23 females and 17 males with a mean age of 47.3±14.3 years. Of those, 33 (82.5 %) were positive for lupus anticoagulants and 25 (62.5 %) were Btriple^ positive for all of the APL tests (i.e. LA, ACL, anti-β2GPI). All of the APS patients were on a long-term anticoagulant therapy of either phenprocoumon (n=28) or warfarin (n=12). Anticoagulant therapy was initiated due to venous thromboembolism in 31 cases (77.5 %), arterial

Ann Hematol

thrombosis in 7 cases (17.5 %) and a mechanical heart valve in 2 cases (5.0 %). The baseline characteristics of the total study cohort are presented in Table 1.

the CFX values were transformed into INR equivalents using the corresponding regression equation for each thromboplastin.

CFX–INR relationship among APL-negative controls

INR and CFX-derived INR equivalent

A goodness-of-fit analysis was performed for the CFX–INR relationship among the APL-negative control subjects (n= 100). Figure 1a shows the scatter plots of the CFX and INR values obtained with different PT assays. Nonlinear regression models revealed an exponential function as the best-fit regression equation: INR=b0 ×CFXb1. The beta coefficients and the coefficient of determination R2 for the test results with the use of the different thromboplastins are presented in Fig. 1a. The best adaptation between the CFX and INR measures was obtained with the use of the Neoplastin Plus reagent (INR= 24.695×CFX−0.698; R2 =0.548). For the subsequent results,

The differences between the INR obtained from PT testing and calculated from the CFX are shown in a Bland–Altman plot (Fig. 1b). The deviation between the INR and CFXderived INR equivalent increased with higher absolute INR values for all of the tested thromboplastins. Upon restricting our analysis to the INR values in the ISI-calibrated INR range of 1.5 to 4.5 and defining an INR difference of ≤0.5 INR units as agreement between two methods, the agreement was best with the Neoplastin Plus reagent in the APL-negative controls (72.0 %) and the Thromborel S reagent in the APS patients (69.1 %) (Table 2). With the use of a laboratory PT assay, there

Table 1 Baseline characteristics of the study cohort Age, years (mean±SD) Males, n (%) Anticoagulant medication Phenprocoumon, n (%) Warfarin, n (%) Duration, years Indication for anticoagulant therapy Venous thromboembolism Arterial thromboembolism Atrial fibrillation Mechanical heart valve Patient’s history Cerebral infarction Myocardial infarction Other systemic embolism Deep venous thrombosis Pulmonary embolism Abortiona Bleeding complications Major bleedingb, % per year Minor bleeding, % per year Comorbidities Chronic kidney diseasec Systemic lupus erythematosus Other autoimmune disorders Malignancies a

Controls, N=100

APS patients, N=40

57.9±17.4 61 (61.0)

47.3±14.3 17 (42.5)

78 (78.0) 22 (22.0) 7.2±9.7

28 (70.0) 12 (30.0) 6.1±6.2

72 (72.0) 3 (3.0) 19 (19.0) 9 (9.0)

31 (77.5) 7 (17.5) 0 (0.0) 2 (5.0)

7 (7.0) 6 (6.0) 4 (4.0) 70 (70.0) 4 (4.0)

11 (27.5) 3 (7.5) 8 (20.0) 28 (70.0) 15 (37.5)

4 (10.3)

7 (30.4)

1.9 3.1

0.8 6.1

16 (16.0) 0 (0.0) 11 (11.0) 13 (13.0)

6 (15.4) 12 (30.0) 1 (2.5) 3 (7.5)

Percentage related to females

b

Defined according to the recommendations of the Scientific and Standardisation Committee of the International Society on Thrombosis and Haemostasis [33] c

Defined according to the standards set by the National Kidney Foundation Disease Outcomes Quality Initiative (NKF NDOQI) (i.e. stages 3 to 5) [34].

Ann Hematol

Fig. 1 a Scatter plots showing the CFX–INR relationship from the INR measurement with different thromboplastins and the best-fit regression curve for this relationship in APL-negative patients (controls; n=100). INR = b 0 × CFX b1 . b Bland–Altman plots showing the absolute

differences of the INR obtained by PT testing and calculated from the CFX values using the best-fit regression equation for each thromboplastin in APS patients and APL-negative patients

was no significant difference in the agreement of the INR and CFX equivalents when comparing the APL-negative controls to either the APS patients or LA-positive APS patients, respectively. There was only a trend towards a lower agreement

with the use of the Neoplastin Plus reagent when comparing the APS patients to the controls (61.7 vs. 72.0 %; p=0.081). However, regarding the results for the point-of-care system CoaguChek XS, an agreement between the INR and the

Ann Hematol Table 2 Agreement between the CFX and INR expressed as the percentage of patients with an INR difference ≤0.5 between the PT-derived INR and CFX-derived INR equivalent. APS patients (N=40) as well as the subgroup of APS patients positive for LA (N= 33) were compared with APLnegative patients (controls)

Controls, N=100

APS patients, N=40

LA-positive patients, N=33

CFX–Neoplastin Plus CFX–RecombiPlasTin CFX–Innovin CFX–Thromborel S

n/N (%) 67 (72.0) 49 (56.3) 46 (51.7) 63 (67.0)

n/N (%) 132 (61.7) 111 (53.4) 103 (51.5) 150 (69.1)

p value 0.081 0.642 0.977 0.714

n/N (%) 108 (62.1) 91 (53.2) 83 (51.2) 118 (66.7)

p value 0.102 0.636 0.945 0.953

CFX–CoaguChek XS

61 (67.8)

114 (55.6)

0.050

92 (55.8)

0.061

APS antiphospholipid syndrome, LA lupus anticoagulants, CFX chromogenic factor X

CFX-derived INR equivalent was less frequently observed in the APS patients (55.6 vs. 67.8 %; p=0.050).

assays in the APS patients. There was also no higher disagreement when we restricted our analysis to LA-positive APS patients.

INR differences between different PT assays Table 3 shows the agreement of the INR values (i.e. INR difference ≤0.5) when comparing the INR obtained with the different PT assays. The best agreement was obtained between RecombiPlasTin and Innovin in APL-negative patients (98.9 %) and between Neoplastin Plus and Thromborel S in APS patients (96.3 %). Significant differences between the APS patients and APL-negative patients were observed when the INR values obtained with the CoaguChek XS system were compared with the results of the laboratory PT assays (i.e. Neoplastin Plus, RecombiPlasTin, Innovin). The agreement between the CoaguChek XS system and Neoplastin Plus was lower for the APS patients compared with the controls (81.5 vs. 91.4 %; p0.5 INR units) obtained with different testing methods is a frequent phenomenon in anticoagulated APS patients as well as in APL-negative patients. When considering all 3058 pairs of INR tests, we did not observe an increased variability of INR values among the APS patients compared with the APLnegative controls. Earlier studies reported significant discrepancies in the INR values measured with different thromboplastins, suggesting that a phospholipid-independent test such as a CFX assay would improve monitoring of VKA therapy in APS patients whose INR values might be affected by the presence of LA. McGlasson et al. analysed the CFX–INR relationship in a

Table 3 Agreement of the INR test results obtained from different PT assays (i.e. deviation ≤0.5 INR units between two methods) comparing all APS patients and APS patients positive for lupus anticoagulants (LA) with APL-negative patients (controls) Paired INRs

Controls (N=93)

APS patients (N=40)

LA-positive patients (N=33)

Neoplastin Plus–RecombiPlasTin Neoplastin Plus–Innovin Neoplastin Plus–Thromborel S Neoplastin Plus–CoaguChek XS

n 307 307 307 304

n/N 63/93 59/93 78/93 85/93

% 67.7 63.4 83.9 91.4

n/N 153/214 150/214 206/214 172/211

% 71.5 70.1 96.3 81.5

p value 0.508 0.251

Monitoring anticoagulant therapy with vitamin K antagonists in patients with antiphospholipid syndrome.

Because of the possible interference of antiphospholipid antibodies (APL) with the phospholipid component of thromboplastin reagents, concerns have be...
624KB Sizes 0 Downloads 8 Views