232 Original article
Monitoring anticoagulation in patients with an unreliable prothrombin time/international normalized ratio: factor II versus chromogenic factor X testing Lisa M. Baumann Kreuzigera, Yvonne H. Dattaa, Andrew D. Johnsonb, Nicole D. Zantekb, Ryan Shanleyc and Mark T. Redinga The international normalized ratio (INR) can be unreliable in patients with lupus anticoagulants (LACs) or other conditions affecting baseline testing. Alternative methods to assess anticoagulation on warfarin through measures of vitamin K-dependent factor activity by clot based or chromogenic assays may be necessary. In this patient population, the ideal method is unknown. Thirty-six patients stable on warfarin with LAC or unreliable INR testing had an INR, a prothrombin time-based clotting assay for factor II (FII) activity, and a chromogenic assay for factor X (CFX) activity were performed simultaneously. Eighty-nine sets of measurements were obtained of which 83 sets included all three assays. CFX and FII levels were well correlated (r U 0.92) in all patients and in 26 patients with a documented antiphospholipid antibody (r U 0.93). Parallel testing was seen in 99% of FII assays. Sixty-one percent of CFX and 57% of FII were within the therapeutic range. In 32 CFX and FII pairs wherein assessment of anticoagulation was discordant, 16 CFX agreed with INR and 13 FII agreed with INR (McNemar’s, x2 U 0.14, P U 0.7). The number of
Introduction The international normalized ratio (INR), a normalized prothrombin time (PT) ratio, is used to measure the anticoagulant effects of warfarin. The vitamin K-dependent factors can be depressed to a variable degree, and factors II (FII) and X are the major determinants of antithrombotic effects in animal models [1]. Human studies have noted that the INR may not accurately reflect the FII level [2]. Additionally, the INR can be an unreliable measure of anticoagulation in patients with antiphospholipid antibodies or other conditions affecting the PT in a nonvitamin K-dependent fashion. Binding of the phospholipid by antiphospholipid antibodies can falsely prolong the PT in patients with lupus anticoagulants (LACs) [3–6]. The degree to which the INR is altered depends on the sensitivity of the thromboplastin or machine calibration to allow an instrument-specific international sensitivity index [3,7,8]. Small subsets of patients with LAC have been reported to still have elevated PTs despite meticulous collection and standardization of testing [7]. Patients with the antiphospholipid antibody syndrome (APS) are at high risk of arterial and venous thromboembolic events and require anticoagulation with warfarin usually for the duration of their lives [9]. In patients with liver disease, reduction in all clotting 0957-5235 ß 2014 Wolters Kluwer Health | Lippincott Williams & Wilkins
times tests were discrepant was not statistically different between CFX and FII (P U 0.36). CFX and FII activity are well correlated in patients that require alternative monitoring of warfarin. Either test can be used in this population. Blood Coagul Fibrinolysis 25:232–236 ß 2014 Wolters Kluwer Health | Lippincott Williams & Wilkins.
Blood Coagulation and Fibrinolysis 2014, 25:232–236 Keywords: anticoagulation, antiphospholipid antibody syndrome, chromogenic factor X, factor II, warfarin a
Division of Hematology, Oncology, and Transplantation, University of Minnesota, Department of Laboratory Medicine and Pathology, University of Minnesota and CTSI Biostatstics, University of Minnesota, Minneapolis, Minnesota, USA
b c
Correspondence to Mark T. Reding, MD, University of Minnesota, Mayo Mail Code 480, 420 Delaware St. S.E., Minneapolis, MN 55455, USA Tel: +1 612 624 0123; fax: +1 612 625 6919; e-mail: [email protected] Received 4 April 2013 Revised 8 October 2013 Accepted 13 November 2013
proteins except for factor VIII leads to prolongation of the PT/INR; however, an increased incidence of thrombosis is also seen in patients with chronic liver disease [10,11]. For patients in whom the PT/INR is falsely prolonged, an alternate way to monitor anticoagulation is required. Rather than measure the degree of anticoagulation by the screening assay INR, vitamin K antagonists can alternatively be monitored by the activity of vitamin K-dependent factors. Activity can be measured through clot based or chromogenic assays. Two alternative methods for monitoring are FII activity and chromogenic factor X (CFX) assays. However, few reports have examined the variability or reliability of these tests in this situation. Rosborough et al. [12] have suggested that CFX assays are preferred over FII testing because two thirds of the patients with LAC had FII/CFX ratios that were less than the median ratio of patients without LAC. The therapeutic range of CFX is 20–40% as established by the literature [3,13], whereas in our laboratory therapeutic, FII is 15–25% [2]. The FII/CFX ratios at the ends of these therapeutic ranges are 0.75 and 0.625. Therefore, FII/CFX ratios less than 1 may not be clinically meaningful because they can occur when both tests are in the therapeutic range. We completed a prospective DOI:10.1097/MBC.0000000000000030
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Factor II versus chromogenic X monitoring Kreuziger et al. 233
cohort study comparing the FII and CFX activities in patients on chronic warfarin with unreliable INR testing to determine the correlation between these tests and reliability based on agreement of therapeutic anticoagulation.
McNemar’s test compared the discordant pairs. Lastly, a mean ratio of FII:CFX was calculated for each patient. Ratios of FII:CFX were compared between patients with and without LAC using the Wilcoxon rank-sum test. The study was approved by the University of Minnesota Institutional Review Board.
Methods Patients on extended warfarin therapy (>6 months) whose anticoagulation was monitored by FII or CFX, termed alternative monitoring group, were identified. The need for alternate warfarin monitoring was determined by individual clinicians without set criteria. Denoted indications included the presence of LAC with baseline PT above the reference range, significant variability of INR testing when nonadhearance was excluded, or documented recurrent thrombosis with INR levels between 2 and 3. Demographic variables and presence of LAC, anticardiolipin or b-2 glycoprotein I antibodies, and determination of documented APS [14] were determined through chart review. Our laboratory performs LAC testing using two of the following methods depending on baseline INR: partial thromboplastin time (PTT) activated with PTT-LA (Diagnostica Stago, Parsippany, New Jersey, USA) or aluminum silicate and rabbit brain thromboplastin and confirmed using platelet neutralization; dilute Russell’s viper venom time (DRVVT) using TriniCLOT Lupus Screen and TriniCLOT Lupus Confirm (Tcoag US Inc., Parsippany New Jersey, USA); and Staclot LA (Diagnostica Stago). Mixing studies were performed with pooled normal plasma obtained either from a commercial source (Precision Biologic, Dartmouth, Nova Socita, Canada) or prepared in-house. Testing was performed on a STA-R or STA-R Evolution (Diagnostica Stago). Anticoagulant effect was prospectively measured on a monthly basis or as clinically necessary over a 3-month period using INR (STA Neoplastine CI Plus, Diagnostica Stago,), FII using a PTbased assay (STA Neoplastine CI Plus and STA deficient II plasma, Diagnostica Stago), and CFX (DiaPharma Factor X kit, DiaPharma; West Chester, Ohio, USA) on a STA-R Evolution (Diagnostica Stago). The CFX assay was performed with HEPES buffered saline (0.05 HEPES/0.15 mol/l NaCl) (Sigma Aldrich, St. Louis, Missouri, USA) or Owren Koller Buffer (Diagnostica Stago) at a 1 : 10 dilution. Assays were performed on the same plasma sample, and at least three dilutions were used in FII assay. Correlation was used to compare INR, FII, and CFX levels. A random effects model was created to account for the repeated measurements in the data set. Therapeutic levels were defined as INR 2.0–3.0, FII 15–25%, and CXF 20–40% [3,13]. The FII therapeutic range was additionally verified in 38 serum samples from patients without LAC or liver disease. In the alternative monitoring group, assays were considered concordant if both FII and CXF were either therapeutic or nontherapeutic.
Results Eighty-nine sets of FII and CFX measurements were obtained in 36 patients over a 3-month timeframe. Eighty-three sets also had an INR evaluated on the same sample. Mean age of participants was 51 years, and 75% were women (Table 1) [14]. Seventy-two percent of patients (26 of 36) had a documented antiphospholipid antibody at our institution, and one additional patient had a reported antiphospholipid antibody in a physician’s note. Five patients (14%) had a positive LAC tested within 1 year, and nine patients (25%) had positive LAC tested within 2 years of the study. Twenty patients (56%) fulfilled the current definition of APS [14]. Of the APS patients, 90% (18 of 20) had positive LAC, and 10% (two of 20) had positive high-titer anticardiolipin antibodies tested at least 12 weeks apart. Additionally, 60% (12 of 20) had high-titer anticardiolipin antibodies, and 55% (11 of 20) had high-titer b-2 glycoprotein 1 antibodies. Twelve patients had two of three tests positive, and nine had positive LAC and high-titer anticardiolipin and b-2 glycoprotein 1 antibodies. Of the six patients with documented antiphospholipid antibodies without confirmed APS, two had one positive LAC with negative repeat testing, one had testing once in our system, one had thrombosis 3 years prior to persistently positive LAC, and two had positive LAC but required anticoagulation due to a mechanical valve instead of thrombosis. CFX and FII levels were well correlated (r ¼ 0.92) in all patients, in 26 patients with a documented antiphospholipid antibody (r ¼ 0.93), and in the 20 patients with APS (r ¼ 0.93). A weaker linear correlation was noted for CFX and INR (r ¼ 0.60) and FII and INR (r ¼ 0.62). Accounting for repeated measures did not significantly Table 1
Baseline demographics and lupus anticoagulant testing n (%) n ¼ 36
Male Age mean (SE) Indication for alternate warfarin monitoring Documented antiphospholipid antibody Antiphospholipid antibody syndrome Reported antiphospholipid antibody Unreliable INR without antiphospholipid antibody Recurrent thrombosis Variable INR
9 (25) 51 (2.5) 26 20 1 9 5 4
(72) (56) (3) (25) (14) (11)
Documented antiphospholipid antibodies include a positive lupus anticoagulant or anticardiolipin IgG or IgM or b-2-glycoprotein I antibody above normal range. Classification of antiphospholipid antibody syndrome based upon international criteria [14]. Reported antiphospholipid antibody category includes patients with documentation in medical record without laboratory testing at our institution. IgG, immunoglobulin G; IgM, immunoglobulin M; INR, international normalized ratio; SE, standard error.
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
234 Blood Coagulation and Fibrinolysis 2014, Vol 25 No 3
Fig. 1
60 40 20
Factor II (%)
80
100
is based on published observations [3,13]. The overall correlation of the INR and FII using therapeutic ranges for INR of 2.0–3.0 and FII of 15–25% was 66% (25 of 38) and 15–30% was 71% (27 of 38). Further widening of the FII therapeutic range did not significantly change the correlation to INR.
1
2
3
4
5
6
7
INR Relationship between factor II and INR. Best fit model: LN(Factor II) ¼ 4.06 – 0.97LN(INR) R-square ¼ 0.84. INR, international normalized ratio.
change the correlation estimates. Figure 1 shows the relationship between FII and INR. The best fit model was LN(Factor II) ¼ 4.06 – 0.97LN(INR) [r ¼ 0.84 (0.76, 0.89)]. Figure 2 shows the relationship between CFX and INR. Similarly, the best fit model with natural log transformation of both variables showed an inverse relationship [LN(X) ¼ 4.13 – 0.90LN(INR) R ¼ 0.83 (0.76, 0.89)]. In a separate set of 38 samples from 12 patients on warfarin without LAC or liver disease, FII levels were compared with INR testing on the same plasma sample to evaluate the FII therapeutic range (Fig. 3). CFX were not obtained on these samples, as the CFX therapeutic range
60 20
40
CFX (%)
80
100
Fig. 2
1
2
3
4
5
6
7
INR Relationship between chromogenic X and INR. Best fit model: LN(X) ¼ 4.13 – 0.90LN(INR) R-square ¼ 0.83. INR, international normalized ratio.
When the alternative monitoring group samples were evaluated using different FII therapeutic ranges, 57% of FII levels were between 15 and 25%, and 69% were between 15 and 30% (Table 2). FII/CFX concordance increased from 61% (54 of 89) to 67% (60 of 89) when the therapeutic range was increased from 15–25% to 15– 30%. There were 10 pairs that differed when FII/CFX concordance was compared between the FII therapeutic levels of 15–25% and 15–30%. On the basis of a therapeutic range of 15–25%, seven of 10 measurements lead to warfarin dose changes that corresponded to therapeutic anticoagulation on the next measurement. These warfarin dose adjustments would not have occurred if a therapeutic range of 15–30% would have been used. In two instances, warfarin dose adjustment did not occur based on a therapeutic range of 15–25%, and the patient required a dose adjustment once on the next measurement. Lastly, one patient’s warfarin dose was not adjusted because a missed dose accounted for the subtherapeutic anticoagulation. Overall, these data support the use of an FII therapeutic range of 15–25%. In 89 sets of tests, 61% of CFX and 57% of FII results were within the therapeutic range. The INR was between 2.0 and 3.0 in 42% (35 of 83) and above 3.0 in 47% (39 of 83) tests. Of the 51 pairs with FII/CFX concordant testing and PT/INR data, the INR did not agree in 24% of the measurements. In all of these instances, the INR showed increased anticoagulant effect in comparison to the FII or CFX levels. This supports the suggestion that the INR was not a reliable indicator of these patients’ anticoagulation status. Thirty-two discordant CFX and FII pairs were found in which an INR was also completed (Table 2). In three samples, none of the testing agreed as the CFX showed therapeutic, FII showed subtherapeutic, and INR showed supratherapeutic anticoagulation. In 16 instances, CFX agreed with INR and in 13 samples, FII agreed with INR (McNemar’s, x2 ¼ 0.14, P ¼ 0.7). Discordant testing was not associated with the presence of an LAC (chi-square P ¼ 0.76). On average, CFX was discrepant by 2.5% activity (range 1–8%). When CFX and INR agreed, FII was higher than expected by an average of 3.2% activity (range 1–8%). The amount that the tests were discrepant was not statistically different between CFX and FII (t-test P ¼ 0.36). Almost all of the patients with a history of LAC had testing completed at our institution with individual tests available. Of the 26 patients, 84% had a prolonged PTT without correction on a mixing study, 54% confirmed
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Factor II versus chromogenic X monitoring Kreuziger et al. 235
Fig. 3
70 60
Factor II (%)
50 40 30 20 10 0 0.00
1.00
2.00
3.00
4.00
5.00
6.00
7.00
INR Relationship between factor II and INR in patients without a lupus anticoagulant or liver disease. INR, international normalized ratio.
with the Staclot LA method, and 28% confirmed by the DRVVT method. In patients with discordant FII and CFX testing, only 31% had a positive LAC within 2 years of the study, and 29% had negative testing. No association was seen between the presence of an LAC and repeat discordant testing. The average FII/CFX ratio was similar between patients with LAC (median 0.89, range 0.51–1.38) and those without LAC (median 0.87, range 0.45–1.27; Wilcoxon rank-sum P ¼ 0.63). Only 1% (one of 89) tests had suggestion of nonparallelism between the 1 : 10 and 1 : 40 dilution with a FII level increase from 88 to 105%. The addition of a 1 : 80 dilution allowed accurate calculation of the FII level. The other 88 FII assays had a mean absolute difference of 0.4 (range 3–3) between the 1 : 10 and 1 : 40 dilutions. Overall, the history of an LAC did not significantly affect the FII or CFX testing.
Discussion In a subset of patients, the PT/INR does not accurately reflect the anticoagulation status on warfarin. A gold standard for monitoring anticoagulation in these patients Table 2
has not been established. FII levels are measured in a clot-based assay, which could theoretically be influenced by interfering antibodies. However, the use of multiple dilutions (typically three) limits the influence of an inhibitor on the factor level determination, and only one sample suggested nonparallelism that was overcome with an additional dilution. The CFX assay requires minimal phospholipid, and the initial dilution is large, which would make the assay less sensitive to an LAC. However, CFX is affected by hyperbilirubinemia, hyperlipidemia, and hemolysis [15]. Both tests have benefits and limitations and require knowledge of the clinical information in order to accurately interpret test results. Our study showed that CFX and FII activity are well correlated in patients in whom alternate anticoagulation monitoring is clinically necessary. Similar correlations were seen between CFX and INR and FII and INR. The best fit model of the relationship between CFX and INR was inverse with a natural log transformation of both variables. Modeling a quadratic relationship with reciprocal transformation of INR, as has been reported in
Factor II and chromogenic factor X testing concordance with varying factor II therapeutic ranges Factor II therapeutic range 15–25% [test number n (%)]
FII tests within therapeutic range FII/CFX concordant pairs FII/CFX discordant pairs FII agreed PT/INR [n (% of discordant pairs)] CFX agreed PT/INR [n (% of discordant pairs)] Neither agreed PT/INR [n (% of discordant pairs)] PT/INR not done [n (% of discordant pairs)] Three test concordance
51 54 35 13 16 3 3 39
(57%) (61%) (39%) (37%) (46%) (8.5%) (8.5%) (45%)
15–30% [test number n (%)] 61 60 29 14 12 1 2 43
(69%) (67%) (33%) (48%) (41%) (4%) (7%) (51%)
CFX, chromogenic factor X; FII, factor II; PT/INR, prothrombin time/international normalized ratio. Testing considered concordant if FII and CFX were both within, both above, or both below the defined therapeutic range.
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
236 Blood Coagulation and Fibrinolysis 2014, Vol 25 No 3
previous studies [16] showed increased sensitivity to outliers in our data. This may be because we had fewer data points at the extreme INR values, and our patients were on stable warfarin dosing. Rosborough et al. [12] have suggested that CFX assays are preferred over FII testing because the FII/CFX ratios were higher in patients without LAC compared with patients with LAC. In our data set, FII/CFX ratios were similar to their data; however, the FII/CFX ratios did not differ between patients with and without LAC [12]. The variability in FII/CFX ratio was less in patients without LAC, but there was insufficient evidence to suggest a difference in the ratios between patients with and without LAC. The discrepant results in comparison to previous studies may have been because of differences in patient selection, as our study enrolled patients whose physicians believed alternate monitoring of warfarin was required in comparison to a chronic warfarin treatment group [12]. Because of differences in therapeutic ranges between FII and CFX, a low FII/CFX ratio is not clinically meaningful. Additionally, a change in the FII/CFX ratio only suggests that the tests fluctuate relative to each other but does not indicate which test causes the altered ratio. Because of these issues, the FII/CFX ratio was not used to assess either of the tests’ reliability. In an effort to determine if variation in testing was clinically meaningful, we classified testing according to therapeutic ranges. Tests were considered concordant if both FII and CFX were therapeutic or not therapeutic. Increasing the FII therapeutic range from 15–25% to 15– 30% increased the FII/CFX concordance; however, this change would have prevented several warfarin dose changes that were clinically appropriate on follow-up. Additionally, as patients with APS are at high risk of thrombosis, tighter control of anticoagulation and more limited therapeutic range is reasonable. We would anticipate if an LAC was interfering with testing that the INR and FII would show an increased anticoagulant effect of warfarin, and the discordant pairs would have shown more agreement of FII with INR. On the contrary, discordant testing between FII and CFX showed approximately equal number of tests with CFX and INR and FII and INR agreement. Therefore, we do not have evidence that either CFX or FII was the preferred alternative monitoring method. Overall, we found that FII and CFX testing are well correlated in patients in whom anticoagulation with warfarin requires alternative monitoring to INR. Either test can be used to measure anticoagulation in this population.
Acknowledgements The authors would like to thank Rachael Rivard for statistical assistance. Funding provided by University of Minnesota Division of Hematology, Oncology and Transplant and NIH T32 Training Grant for LBK research time. Conflicts of interest
There are no conflicts of interest related to content of the article.
References 1
2
3 4
5
6 7
8
9
10
11
12
13
14
15
16
Zivelin A, Rao LV, Rapaport SI. Mechanism of the anticoagulant effect of warfarin as evaluated in rabbits by selective depression of individual procoagulant vitamin K-dependent clotting factors. J Clin Invest 1993; 92:2131–2140. Lind SE, Callas PW, Golden EA, Joyner KA, Ortel TL. Plasma levels of factors II, VII and X and their relationship to the international normalized ratio during chronic warfarin therapy. Blood Coagul Fibrinolysis 1997; 8:48– 53. Moll S, Ortel TL. Monitoring warfarin therapy in patients with lupus anticoagulants. Ann Intern Med 1997; 127:177–185. Gastineau DA, Kazmier FJ, Nichols WL, Bowie EJ. Lupus anticoagulant: an analysis of the clinical and laboratory features of 219 cases. Am J Hematol 1985; 19:265–275. Horellou MH, Aurousseau MH, Boffa MC, Conrad J, Wiesel ML, Samama M. Biological and clinical heterogeneity of lupus and lupus-like anticoagulant in fifty-seven patients. J Med 1987; 18:199–217. Fleck RA, Rapaport SI, Rao LV. Antiprothrombin antibodies and the lupus anticoagulant. Blood 1988; 72:512–519. Robert A, Le Querrec A, Delahousse B, Caron C, Houbouyan L, Boutire B, et al. Control of oral anticoagulation in patients with the antiphospholipid syndrome: influence of the lupus anticoagulant on International Normalized Ratio. Groupe Me´thodologie en He´mostase du Groupe d’Etudes sur l’He´mostases et la Thrombose. Thromb Haemost 1998; 80:99–103. Tripodi A, Chantarangkul V, Clerici M, Negri B, Galli M, Mannucci PM. Laboratory control of oral anticoagulant treatment by the INR system in patients with the antiphospholipid syndrome and lupus anticoagulant. Results of a collaborative study involving nine commercial thromboplastins. Br J Haematol 2001; 115:672–678. Khamashta MA, Cuadrado MJ, Mujic F, Taub NA, Hunt BJ, Hughes GR. The management of thrombosis in the antiphospholipid-antibody syndrome. N Engl J Med 1995; 332:993–997. Kovacs MJ, Wong A, MacKinnon K, Weir K, Keeney M, Boyel E, et al. Assessment of the validity of the INR system for patients with liver impairment. Thromb Haemost 1994; 71:727–730. Pincus K, Tata A, Watson K. Risk of venous thromboembolism in patients with chronic liver disease and the utility of venous thromboembolism prophylaxis. Ann Pharmacother 2012; 46:873–878. Rosborough T, Jacobsen J, Shepherd M. Factor X and factor II activity levels do not always agree in warfarin-treated lupus anticoagulant patients. Blood Coagul Fibrinolysis 2010; 21:242–244. McGlasson DL, Romick BG, Rubal BJ. Comparison of a chromogenic factor X assay with international normalized ratio for monitoring oral anticoagulation therapy. Blood Coagul Fibrinolysis 2008; 19:513–517. DiaPharma. DiaPharma Factor X package insert. http:// www.diapharma.com/uploads/insert/KDPGFXinsertR2_0606.pdf . Revision6/2006. [Accessed on 20 November 2013]. Miyakis S, Lockshin MD, Atsumi T, Branch DW, Brey RL, Cervera R, et al. International consensus statement on an update of the classification criteria for definite antiphospholipid syndrome (APS). J Thromb Haemost 2006; 4:295–306. Rosborough T, Jacobsen J, Shepherd M. Relationship between chromogenic factor X and international normalized ratio differs during early warfarin initiation compared with chronic warfarin administration. Blood Coagul Fibrinolysis 2009; 20:433–435.
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Monitoring anticoagulation in patients with an unreliable prothrombin time/international normalized ratio: factor II versus chromogenic factor X testing Lisa M. Baumann Kreuzigera, Yvonne H. Dattaa, Andrew D. Johnsonb, Nicole D. Zantekb, Ryan Shanleyc and Mark T. Redinga The international normalized ratio (INR) can be unreliable in patients with lupus anticoagulants (LACs) or other conditions affecting baseline testing. Alternative methods to assess anticoagulation on warfarin through measures of vitamin K-dependent factor activity by clot based or chromogenic assays may be necessary. In this patient population, the ideal method is unknown. Thirty-six patients stable on warfarin with LAC or unreliable INR testing had an INR, a prothrombin time-based clotting assay for factor II (FII) activity, and a chromogenic assay for factor X (CFX) activity were performed simultaneously. Eighty-nine sets of measurements were obtained of which 83 sets included all three assays. CFX and FII levels were well correlated (r U 0.92) in all patients and in 26 patients with a documented antiphospholipid antibody (r U 0.93). Parallel testing was seen in 99% of FII assays. Sixty-one percent of CFX and 57% of FII were within the therapeutic range. In 32 CFX and FII pairs wherein assessment of anticoagulation was discordant, 16 CFX agreed with INR and 13 FII agreed with INR (McNemar’s, x2 U 0.14, P U 0.7). The number of
Introduction The international normalized ratio (INR), a normalized prothrombin time (PT) ratio, is used to measure the anticoagulant effects of warfarin. The vitamin K-dependent factors can be depressed to a variable degree, and factors II (FII) and X are the major determinants of antithrombotic effects in animal models [1]. Human studies have noted that the INR may not accurately reflect the FII level [2]. Additionally, the INR can be an unreliable measure of anticoagulation in patients with antiphospholipid antibodies or other conditions affecting the PT in a nonvitamin K-dependent fashion. Binding of the phospholipid by antiphospholipid antibodies can falsely prolong the PT in patients with lupus anticoagulants (LACs) [3–6]. The degree to which the INR is altered depends on the sensitivity of the thromboplastin or machine calibration to allow an instrument-specific international sensitivity index [3,7,8]. Small subsets of patients with LAC have been reported to still have elevated PTs despite meticulous collection and standardization of testing [7]. Patients with the antiphospholipid antibody syndrome (APS) are at high risk of arterial and venous thromboembolic events and require anticoagulation with warfarin usually for the duration of their lives [9]. In patients with liver disease, reduction in all clotting 0957-5235 ß 2014 Wolters Kluwer Health | Lippincott Williams & Wilkins
times tests were discrepant was not statistically different between CFX and FII (P U 0.36). CFX and FII activity are well correlated in patients that require alternative monitoring of warfarin. Either test can be used in this population. Blood Coagul Fibrinolysis 25:232–236 ß 2014 Wolters Kluwer Health | Lippincott Williams & Wilkins.
Blood Coagulation and Fibrinolysis 2014, 25:232–236 Keywords: anticoagulation, antiphospholipid antibody syndrome, chromogenic factor X, factor II, warfarin a
Division of Hematology, Oncology, and Transplantation, University of Minnesota, Department of Laboratory Medicine and Pathology, University of Minnesota and CTSI Biostatstics, University of Minnesota, Minneapolis, Minnesota, USA
b c
Correspondence to Mark T. Reding, MD, University of Minnesota, Mayo Mail Code 480, 420 Delaware St. S.E., Minneapolis, MN 55455, USA Tel: +1 612 624 0123; fax: +1 612 625 6919; e-mail: [email protected] Received 4 April 2013 Revised 8 October 2013 Accepted 13 November 2013
proteins except for factor VIII leads to prolongation of the PT/INR; however, an increased incidence of thrombosis is also seen in patients with chronic liver disease [10,11]. For patients in whom the PT/INR is falsely prolonged, an alternate way to monitor anticoagulation is required. Rather than measure the degree of anticoagulation by the screening assay INR, vitamin K antagonists can alternatively be monitored by the activity of vitamin K-dependent factors. Activity can be measured through clot based or chromogenic assays. Two alternative methods for monitoring are FII activity and chromogenic factor X (CFX) assays. However, few reports have examined the variability or reliability of these tests in this situation. Rosborough et al. [12] have suggested that CFX assays are preferred over FII testing because two thirds of the patients with LAC had FII/CFX ratios that were less than the median ratio of patients without LAC. The therapeutic range of CFX is 20–40% as established by the literature [3,13], whereas in our laboratory therapeutic, FII is 15–25% [2]. The FII/CFX ratios at the ends of these therapeutic ranges are 0.75 and 0.625. Therefore, FII/CFX ratios less than 1 may not be clinically meaningful because they can occur when both tests are in the therapeutic range. We completed a prospective DOI:10.1097/MBC.0000000000000030
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Factor II versus chromogenic X monitoring Kreuziger et al. 233
cohort study comparing the FII and CFX activities in patients on chronic warfarin with unreliable INR testing to determine the correlation between these tests and reliability based on agreement of therapeutic anticoagulation.
McNemar’s test compared the discordant pairs. Lastly, a mean ratio of FII:CFX was calculated for each patient. Ratios of FII:CFX were compared between patients with and without LAC using the Wilcoxon rank-sum test. The study was approved by the University of Minnesota Institutional Review Board.
Methods Patients on extended warfarin therapy (>6 months) whose anticoagulation was monitored by FII or CFX, termed alternative monitoring group, were identified. The need for alternate warfarin monitoring was determined by individual clinicians without set criteria. Denoted indications included the presence of LAC with baseline PT above the reference range, significant variability of INR testing when nonadhearance was excluded, or documented recurrent thrombosis with INR levels between 2 and 3. Demographic variables and presence of LAC, anticardiolipin or b-2 glycoprotein I antibodies, and determination of documented APS [14] were determined through chart review. Our laboratory performs LAC testing using two of the following methods depending on baseline INR: partial thromboplastin time (PTT) activated with PTT-LA (Diagnostica Stago, Parsippany, New Jersey, USA) or aluminum silicate and rabbit brain thromboplastin and confirmed using platelet neutralization; dilute Russell’s viper venom time (DRVVT) using TriniCLOT Lupus Screen and TriniCLOT Lupus Confirm (Tcoag US Inc., Parsippany New Jersey, USA); and Staclot LA (Diagnostica Stago). Mixing studies were performed with pooled normal plasma obtained either from a commercial source (Precision Biologic, Dartmouth, Nova Socita, Canada) or prepared in-house. Testing was performed on a STA-R or STA-R Evolution (Diagnostica Stago). Anticoagulant effect was prospectively measured on a monthly basis or as clinically necessary over a 3-month period using INR (STA Neoplastine CI Plus, Diagnostica Stago,), FII using a PTbased assay (STA Neoplastine CI Plus and STA deficient II plasma, Diagnostica Stago), and CFX (DiaPharma Factor X kit, DiaPharma; West Chester, Ohio, USA) on a STA-R Evolution (Diagnostica Stago). The CFX assay was performed with HEPES buffered saline (0.05 HEPES/0.15 mol/l NaCl) (Sigma Aldrich, St. Louis, Missouri, USA) or Owren Koller Buffer (Diagnostica Stago) at a 1 : 10 dilution. Assays were performed on the same plasma sample, and at least three dilutions were used in FII assay. Correlation was used to compare INR, FII, and CFX levels. A random effects model was created to account for the repeated measurements in the data set. Therapeutic levels were defined as INR 2.0–3.0, FII 15–25%, and CXF 20–40% [3,13]. The FII therapeutic range was additionally verified in 38 serum samples from patients without LAC or liver disease. In the alternative monitoring group, assays were considered concordant if both FII and CXF were either therapeutic or nontherapeutic.
Results Eighty-nine sets of FII and CFX measurements were obtained in 36 patients over a 3-month timeframe. Eighty-three sets also had an INR evaluated on the same sample. Mean age of participants was 51 years, and 75% were women (Table 1) [14]. Seventy-two percent of patients (26 of 36) had a documented antiphospholipid antibody at our institution, and one additional patient had a reported antiphospholipid antibody in a physician’s note. Five patients (14%) had a positive LAC tested within 1 year, and nine patients (25%) had positive LAC tested within 2 years of the study. Twenty patients (56%) fulfilled the current definition of APS [14]. Of the APS patients, 90% (18 of 20) had positive LAC, and 10% (two of 20) had positive high-titer anticardiolipin antibodies tested at least 12 weeks apart. Additionally, 60% (12 of 20) had high-titer anticardiolipin antibodies, and 55% (11 of 20) had high-titer b-2 glycoprotein 1 antibodies. Twelve patients had two of three tests positive, and nine had positive LAC and high-titer anticardiolipin and b-2 glycoprotein 1 antibodies. Of the six patients with documented antiphospholipid antibodies without confirmed APS, two had one positive LAC with negative repeat testing, one had testing once in our system, one had thrombosis 3 years prior to persistently positive LAC, and two had positive LAC but required anticoagulation due to a mechanical valve instead of thrombosis. CFX and FII levels were well correlated (r ¼ 0.92) in all patients, in 26 patients with a documented antiphospholipid antibody (r ¼ 0.93), and in the 20 patients with APS (r ¼ 0.93). A weaker linear correlation was noted for CFX and INR (r ¼ 0.60) and FII and INR (r ¼ 0.62). Accounting for repeated measures did not significantly Table 1
Baseline demographics and lupus anticoagulant testing n (%) n ¼ 36
Male Age mean (SE) Indication for alternate warfarin monitoring Documented antiphospholipid antibody Antiphospholipid antibody syndrome Reported antiphospholipid antibody Unreliable INR without antiphospholipid antibody Recurrent thrombosis Variable INR
9 (25) 51 (2.5) 26 20 1 9 5 4
(72) (56) (3) (25) (14) (11)
Documented antiphospholipid antibodies include a positive lupus anticoagulant or anticardiolipin IgG or IgM or b-2-glycoprotein I antibody above normal range. Classification of antiphospholipid antibody syndrome based upon international criteria [14]. Reported antiphospholipid antibody category includes patients with documentation in medical record without laboratory testing at our institution. IgG, immunoglobulin G; IgM, immunoglobulin M; INR, international normalized ratio; SE, standard error.
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
234 Blood Coagulation and Fibrinolysis 2014, Vol 25 No 3
Fig. 1
60 40 20
Factor II (%)
80
100
is based on published observations [3,13]. The overall correlation of the INR and FII using therapeutic ranges for INR of 2.0–3.0 and FII of 15–25% was 66% (25 of 38) and 15–30% was 71% (27 of 38). Further widening of the FII therapeutic range did not significantly change the correlation to INR.
1
2
3
4
5
6
7
INR Relationship between factor II and INR. Best fit model: LN(Factor II) ¼ 4.06 – 0.97LN(INR) R-square ¼ 0.84. INR, international normalized ratio.
change the correlation estimates. Figure 1 shows the relationship between FII and INR. The best fit model was LN(Factor II) ¼ 4.06 – 0.97LN(INR) [r ¼ 0.84 (0.76, 0.89)]. Figure 2 shows the relationship between CFX and INR. Similarly, the best fit model with natural log transformation of both variables showed an inverse relationship [LN(X) ¼ 4.13 – 0.90LN(INR) R ¼ 0.83 (0.76, 0.89)]. In a separate set of 38 samples from 12 patients on warfarin without LAC or liver disease, FII levels were compared with INR testing on the same plasma sample to evaluate the FII therapeutic range (Fig. 3). CFX were not obtained on these samples, as the CFX therapeutic range
60 20
40
CFX (%)
80
100
Fig. 2
1
2
3
4
5
6
7
INR Relationship between chromogenic X and INR. Best fit model: LN(X) ¼ 4.13 – 0.90LN(INR) R-square ¼ 0.83. INR, international normalized ratio.
When the alternative monitoring group samples were evaluated using different FII therapeutic ranges, 57% of FII levels were between 15 and 25%, and 69% were between 15 and 30% (Table 2). FII/CFX concordance increased from 61% (54 of 89) to 67% (60 of 89) when the therapeutic range was increased from 15–25% to 15– 30%. There were 10 pairs that differed when FII/CFX concordance was compared between the FII therapeutic levels of 15–25% and 15–30%. On the basis of a therapeutic range of 15–25%, seven of 10 measurements lead to warfarin dose changes that corresponded to therapeutic anticoagulation on the next measurement. These warfarin dose adjustments would not have occurred if a therapeutic range of 15–30% would have been used. In two instances, warfarin dose adjustment did not occur based on a therapeutic range of 15–25%, and the patient required a dose adjustment once on the next measurement. Lastly, one patient’s warfarin dose was not adjusted because a missed dose accounted for the subtherapeutic anticoagulation. Overall, these data support the use of an FII therapeutic range of 15–25%. In 89 sets of tests, 61% of CFX and 57% of FII results were within the therapeutic range. The INR was between 2.0 and 3.0 in 42% (35 of 83) and above 3.0 in 47% (39 of 83) tests. Of the 51 pairs with FII/CFX concordant testing and PT/INR data, the INR did not agree in 24% of the measurements. In all of these instances, the INR showed increased anticoagulant effect in comparison to the FII or CFX levels. This supports the suggestion that the INR was not a reliable indicator of these patients’ anticoagulation status. Thirty-two discordant CFX and FII pairs were found in which an INR was also completed (Table 2). In three samples, none of the testing agreed as the CFX showed therapeutic, FII showed subtherapeutic, and INR showed supratherapeutic anticoagulation. In 16 instances, CFX agreed with INR and in 13 samples, FII agreed with INR (McNemar’s, x2 ¼ 0.14, P ¼ 0.7). Discordant testing was not associated with the presence of an LAC (chi-square P ¼ 0.76). On average, CFX was discrepant by 2.5% activity (range 1–8%). When CFX and INR agreed, FII was higher than expected by an average of 3.2% activity (range 1–8%). The amount that the tests were discrepant was not statistically different between CFX and FII (t-test P ¼ 0.36). Almost all of the patients with a history of LAC had testing completed at our institution with individual tests available. Of the 26 patients, 84% had a prolonged PTT without correction on a mixing study, 54% confirmed
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Factor II versus chromogenic X monitoring Kreuziger et al. 235
Fig. 3
70 60
Factor II (%)
50 40 30 20 10 0 0.00
1.00
2.00
3.00
4.00
5.00
6.00
7.00
INR Relationship between factor II and INR in patients without a lupus anticoagulant or liver disease. INR, international normalized ratio.
with the Staclot LA method, and 28% confirmed by the DRVVT method. In patients with discordant FII and CFX testing, only 31% had a positive LAC within 2 years of the study, and 29% had negative testing. No association was seen between the presence of an LAC and repeat discordant testing. The average FII/CFX ratio was similar between patients with LAC (median 0.89, range 0.51–1.38) and those without LAC (median 0.87, range 0.45–1.27; Wilcoxon rank-sum P ¼ 0.63). Only 1% (one of 89) tests had suggestion of nonparallelism between the 1 : 10 and 1 : 40 dilution with a FII level increase from 88 to 105%. The addition of a 1 : 80 dilution allowed accurate calculation of the FII level. The other 88 FII assays had a mean absolute difference of 0.4 (range 3–3) between the 1 : 10 and 1 : 40 dilutions. Overall, the history of an LAC did not significantly affect the FII or CFX testing.
Discussion In a subset of patients, the PT/INR does not accurately reflect the anticoagulation status on warfarin. A gold standard for monitoring anticoagulation in these patients Table 2
has not been established. FII levels are measured in a clot-based assay, which could theoretically be influenced by interfering antibodies. However, the use of multiple dilutions (typically three) limits the influence of an inhibitor on the factor level determination, and only one sample suggested nonparallelism that was overcome with an additional dilution. The CFX assay requires minimal phospholipid, and the initial dilution is large, which would make the assay less sensitive to an LAC. However, CFX is affected by hyperbilirubinemia, hyperlipidemia, and hemolysis [15]. Both tests have benefits and limitations and require knowledge of the clinical information in order to accurately interpret test results. Our study showed that CFX and FII activity are well correlated in patients in whom alternate anticoagulation monitoring is clinically necessary. Similar correlations were seen between CFX and INR and FII and INR. The best fit model of the relationship between CFX and INR was inverse with a natural log transformation of both variables. Modeling a quadratic relationship with reciprocal transformation of INR, as has been reported in
Factor II and chromogenic factor X testing concordance with varying factor II therapeutic ranges Factor II therapeutic range 15–25% [test number n (%)]
FII tests within therapeutic range FII/CFX concordant pairs FII/CFX discordant pairs FII agreed PT/INR [n (% of discordant pairs)] CFX agreed PT/INR [n (% of discordant pairs)] Neither agreed PT/INR [n (% of discordant pairs)] PT/INR not done [n (% of discordant pairs)] Three test concordance
51 54 35 13 16 3 3 39
(57%) (61%) (39%) (37%) (46%) (8.5%) (8.5%) (45%)
15–30% [test number n (%)] 61 60 29 14 12 1 2 43
(69%) (67%) (33%) (48%) (41%) (4%) (7%) (51%)
CFX, chromogenic factor X; FII, factor II; PT/INR, prothrombin time/international normalized ratio. Testing considered concordant if FII and CFX were both within, both above, or both below the defined therapeutic range.
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
236 Blood Coagulation and Fibrinolysis 2014, Vol 25 No 3
previous studies [16] showed increased sensitivity to outliers in our data. This may be because we had fewer data points at the extreme INR values, and our patients were on stable warfarin dosing. Rosborough et al. [12] have suggested that CFX assays are preferred over FII testing because the FII/CFX ratios were higher in patients without LAC compared with patients with LAC. In our data set, FII/CFX ratios were similar to their data; however, the FII/CFX ratios did not differ between patients with and without LAC [12]. The variability in FII/CFX ratio was less in patients without LAC, but there was insufficient evidence to suggest a difference in the ratios between patients with and without LAC. The discrepant results in comparison to previous studies may have been because of differences in patient selection, as our study enrolled patients whose physicians believed alternate monitoring of warfarin was required in comparison to a chronic warfarin treatment group [12]. Because of differences in therapeutic ranges between FII and CFX, a low FII/CFX ratio is not clinically meaningful. Additionally, a change in the FII/CFX ratio only suggests that the tests fluctuate relative to each other but does not indicate which test causes the altered ratio. Because of these issues, the FII/CFX ratio was not used to assess either of the tests’ reliability. In an effort to determine if variation in testing was clinically meaningful, we classified testing according to therapeutic ranges. Tests were considered concordant if both FII and CFX were therapeutic or not therapeutic. Increasing the FII therapeutic range from 15–25% to 15– 30% increased the FII/CFX concordance; however, this change would have prevented several warfarin dose changes that were clinically appropriate on follow-up. Additionally, as patients with APS are at high risk of thrombosis, tighter control of anticoagulation and more limited therapeutic range is reasonable. We would anticipate if an LAC was interfering with testing that the INR and FII would show an increased anticoagulant effect of warfarin, and the discordant pairs would have shown more agreement of FII with INR. On the contrary, discordant testing between FII and CFX showed approximately equal number of tests with CFX and INR and FII and INR agreement. Therefore, we do not have evidence that either CFX or FII was the preferred alternative monitoring method. Overall, we found that FII and CFX testing are well correlated in patients in whom anticoagulation with warfarin requires alternative monitoring to INR. Either test can be used to measure anticoagulation in this population.
Acknowledgements The authors would like to thank Rachael Rivard for statistical assistance. Funding provided by University of Minnesota Division of Hematology, Oncology and Transplant and NIH T32 Training Grant for LBK research time. Conflicts of interest
There are no conflicts of interest related to content of the article.
References 1
2
3 4
5
6 7
8
9
10
11
12
13
14
15
16
Zivelin A, Rao LV, Rapaport SI. Mechanism of the anticoagulant effect of warfarin as evaluated in rabbits by selective depression of individual procoagulant vitamin K-dependent clotting factors. J Clin Invest 1993; 92:2131–2140. Lind SE, Callas PW, Golden EA, Joyner KA, Ortel TL. Plasma levels of factors II, VII and X and their relationship to the international normalized ratio during chronic warfarin therapy. Blood Coagul Fibrinolysis 1997; 8:48– 53. Moll S, Ortel TL. Monitoring warfarin therapy in patients with lupus anticoagulants. Ann Intern Med 1997; 127:177–185. Gastineau DA, Kazmier FJ, Nichols WL, Bowie EJ. Lupus anticoagulant: an analysis of the clinical and laboratory features of 219 cases. Am J Hematol 1985; 19:265–275. Horellou MH, Aurousseau MH, Boffa MC, Conrad J, Wiesel ML, Samama M. Biological and clinical heterogeneity of lupus and lupus-like anticoagulant in fifty-seven patients. J Med 1987; 18:199–217. Fleck RA, Rapaport SI, Rao LV. Antiprothrombin antibodies and the lupus anticoagulant. Blood 1988; 72:512–519. Robert A, Le Querrec A, Delahousse B, Caron C, Houbouyan L, Boutire B, et al. Control of oral anticoagulation in patients with the antiphospholipid syndrome: influence of the lupus anticoagulant on International Normalized Ratio. Groupe Me´thodologie en He´mostase du Groupe d’Etudes sur l’He´mostases et la Thrombose. Thromb Haemost 1998; 80:99–103. Tripodi A, Chantarangkul V, Clerici M, Negri B, Galli M, Mannucci PM. Laboratory control of oral anticoagulant treatment by the INR system in patients with the antiphospholipid syndrome and lupus anticoagulant. Results of a collaborative study involving nine commercial thromboplastins. Br J Haematol 2001; 115:672–678. Khamashta MA, Cuadrado MJ, Mujic F, Taub NA, Hunt BJ, Hughes GR. The management of thrombosis in the antiphospholipid-antibody syndrome. N Engl J Med 1995; 332:993–997. Kovacs MJ, Wong A, MacKinnon K, Weir K, Keeney M, Boyel E, et al. Assessment of the validity of the INR system for patients with liver impairment. Thromb Haemost 1994; 71:727–730. Pincus K, Tata A, Watson K. Risk of venous thromboembolism in patients with chronic liver disease and the utility of venous thromboembolism prophylaxis. Ann Pharmacother 2012; 46:873–878. Rosborough T, Jacobsen J, Shepherd M. Factor X and factor II activity levels do not always agree in warfarin-treated lupus anticoagulant patients. Blood Coagul Fibrinolysis 2010; 21:242–244. McGlasson DL, Romick BG, Rubal BJ. Comparison of a chromogenic factor X assay with international normalized ratio for monitoring oral anticoagulation therapy. Blood Coagul Fibrinolysis 2008; 19:513–517. DiaPharma. DiaPharma Factor X package insert. http:// www.diapharma.com/uploads/insert/KDPGFXinsertR2_0606.pdf . Revision6/2006. [Accessed on 20 November 2013]. Miyakis S, Lockshin MD, Atsumi T, Branch DW, Brey RL, Cervera R, et al. International consensus statement on an update of the classification criteria for definite antiphospholipid syndrome (APS). J Thromb Haemost 2006; 4:295–306. Rosborough T, Jacobsen J, Shepherd M. Relationship between chromogenic factor X and international normalized ratio differs during early warfarin initiation compared with chronic warfarin administration. Blood Coagul Fibrinolysis 2009; 20:433–435.
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
