Lupus Anticoagulant Testing Performance and Practices by North American Clinical Laboratories

Coagulation and Transfusion Medicine / Lupus Anticoagulant Testing Lupus Anticoagulant Testing Performance and Practices by North American Clinical Laboratories Francine R. Dembitzer, MD, 1 Marlies R. Ledford Kraemer, MBA, BS, MT(ASCP)SH, 2 Piet Meijer, PhD, 3 and Ellinor I.B. Peerschke, PhD 1 Key Words: Lupus anticoagulant; Proficiency testing; Activated partial thromboplastin time; Dilute Russell viper venom time DOI: 10.1309/AJCP4SPPLG5XVIXF Abstract Lupus anticoagulant (LAC) testing is important for evaluating patients with antiphospholipid syndromes and hypercoagulable states. We reviewed results of proficiency testing challenges (n = 5) distributed by the North American Specialized Coagulation Laboratory Association to examine LAC testing performed by participating laboratories. The activated partial thromboplastin time (APTT) and dilute Russell viper venom time (drvvt) constituted major testing methods. In screening studies, LAC-sensitive APTT methods were more sensitive to weak LAC than drvvt-based methods but less specific. In confirmatory testing, drvvt methods performed better, but performance was LAC-dependent. The highest false-negative confirmatory test results were obtained for the platelet neutralization procedure. Noncompliance with recommendations for LAC testing by the International Society on Thrombosis and Haemostasis was high (8%-38%), with the majority of noncompliant laboratories failing to report results of mixing studies. These data provide new insights into LAC testing in North America and identify opportunities for standardization. Laboratory testing for the presence of a lupus anticoagulant (LAC) is integral to the diagnosis of patients with antiphospholipid syndromes and hypercoagulable states. 1-6 LACs are heterogeneous circulating autoantibodies, predominantly of IgG and IgM isotypes, directed against epitopes found on negatively charged phospholipid-binding proteins, which inhibit phospholipid-dependent coagulation reactions in vitro. 4-7 Paradoxically, except in rare instances, LACs are associated with increased arterial and venous thrombosis, not with bleeding. 2 Clinically, complications of LACs are implicated in stroke, transient ischemic attacks, recurrent spontaneous abortions, and acquired thrombophilia. 2,3,8 Since the thrombotic potential is considerable in patients with LAC, accurate diagnosis is essential for risk assessment and longterm patient management with anticoagulant therapy. 2,9 Significant differences exist among specialized coagulation laboratories with respect to LAC testing assays, methods, practices, and outcomes. 2,3,6,10 Previous studies conducted predominantly in the United Kingdom and Italy demonstrated high false-negative and false-positive rates (~20%) for LAC detection. 1,2 The variability in test results was attributable not only to analytical factors but also to preanalytic and postanalytic considerations. 6,11 To date, no single test has shown 100% sensitivity to LAC. 2 Moreover, there is recognized variability in the sensitivity of commonly used assays, methods, and reagents to different LAC preparations. 1,12 To compound the problem, assays and reagents are variably sensitive also to interferences such as heparin and specific factor deficiencies and inhibitors, leading to false-positive LAC test results. False-negative results have been reported if plasma is not sufficiently platelet poor, and dilutional effects of mixing studies variably impact detection of a weak LAC. 10 764 Am J Clin Pathol 2010;134:764-773 764 DOI: 10.1309/AJCP4SPPLG5XVIXF

Coagulation and Transfusion Medicine / Original Article To improve LAC testing, the International Society of Thrombosis and Haemostasis (ISTH) published testing guidelines in 1995, 13 which were revised in 2009. 14 Recommendations include performing the following: (1) at least 2 screening tests that demonstrate prolongation of a phospholipid-dependent clotting time using different testing principles, (2) a mixing study to confirm the presence of an inhibitor and to exclude a factor deficiency, and (3) a confirmatory test that demonstrates phospholipid-dependent inhibitory activity. The guidelines also suggest ruling out other coagulopathies. Despite existing recommendations, 13 there continues to be a lack of uniformity among laboratories with regard to local testing protocols and procedures and to result interpretation. The present study was designed to evaluate LAC testing performance and practices by North American clinical laboratories, using results from 4 consecutive proficiency testing challenges distributed in 2008 and 1 proficiency testing challenge in 2009. To our knowledge, this is the first analysis specifically focused on clinical laboratories in the United States and Canada. The results provide important insights into LAC testing and identify opportunities for continued efforts at standardization. Materials and Methods The North American Specialized Coagulation Laboratory Association (NASCOLA) is a nonprofit organization that distributes proficiency testing modules to North American clinical laboratories performing diagnostic testing for bleeding and prothrombotic disorders. The organization creates a forum for the critical evaluation of coagulation testing procedures and practices to aid in developing guidelines for appropriate use, performance, and interpretation of coagulation tests and results. The present study focused on LAC testing practices and performance by analyzing results from 4 consecutive proficiency testing surveys distributed in 2008 and 1 in 2009. The number of participating laboratories varied per survey, with 46 to 53 laboratories submitting results. Samples consisted of lyophilized plasma obtained from the ECAT (European Concerted Action on Thrombophilia) Foundation (Leiden, Netherlands). No clinical information was provided. Proficiency testing sample characteristics were as follows: Sample 2008-1 was a commercial pool (lot No. 2V72A00, Technoclone, Vienna, Austria) of high-titer LACpositive plasma samples. Sample 2008-2 was plasma obtained from a single female donor with a medium-titer LAC but no history of thrombotic events. Sample 2008-3 represented a commercial plasma pool (lot No. 2U81A00, Technoclone) with low-titer LAC, prepared from a pool of lupus-positive plasma samples. Sample 2008-4 consisted of a single donor plasma sample with medium-titer LAC obtained from a female patient with no reported history of thrombotic events. This plasma sample was diluted with normal pooled plasma in a ratio of 4:1. Finally, sample 2009-2 represented a normal plasma pool containing no LAC. Participating laboratories were asked to analyze each proficiency testing sample according to their local LAC testing protocol. Each laboratory reported results for screening, mixing, and confirmatory tests and included an overall assessment of the presence or absence of LAC. Results for assay and method combinations reported by 3 or more participants were included in the analysis. Results for improbable assay and method combinations, eg, kaolin recalcification time performed with an activated partial thromboplastin time (APTT) reagent containing ellagic acid, representing postanalytic error, were excluded. Only results reported by participants as clotting times in seconds were evaluated. Isolated laboratories exclusively reported clotting time ratios relative to reference plasma for screening, mixing, and confirmatory testing. The mean and standard deviation (SD) were calculated for numeric data. Screening test results were compared with results obtained with local reference plasma samples by using an unpaired Student t test. A P value of.05 or less was considered statistically significant. For purposes of this study, results of mixing studies were compared by calculating the index of circulating anticoagulant (ICA), also known as the Rosner Index, 15 based on data provided by participants. In the absence of knowing individual laboratory cutoffs, an ICA greater than 15 was considered indicative of an LAC, as originally proposed by Rosner et al. 15 Confirmatory test results were evaluated according to participant interpretation: positive, negative, or borderline positive for LAC. Overall assay performance was evaluated by comparing false-positive and false-negative rates. Finally, compliance with LAC testing guidelines 13,14 was assessed based on result reporting patterns. The data provided insight into the number and type of screening tests performed and compliance with mixing and confirmatory study recommendations. In addition, the impact of compliance with LAC testing guidelines on overall accuracy of final result interpretations was examined. Results Results of 248 LAC testing panels were evaluated. These were submitted by 46 to 53 laboratories for 5 proficiency testing challenges. Samples containing strong (2008-1) and weak (2008-2, 2008-3, and 2008-4) LAC and a normal plasma sample (2009-2) were analyzed. LAC testing was performed using a variety of assay and method combinations. The most Am J Clin Pathol 2010;134:764-773 765 765 DOI: 10.1309/AJCP4SPPLG5XVIXF 765

Dembitzer et al / Lupus Anticoagulant Testing frequently used combinations are listed in Table 1 and represent automated methods. Owing to the large number of assay types and methods relative to the number of reporting laboratories, further subanalysis of data by instrumentation did not have sufficient statistical power. Major screening test results for proficiency challenges considered in the present study are summarized in Table 2. Results were compared by assay and method combinations. Participating NASCOLA laboratories performed predominantly APTT- and/or dilute Russell viper venom time (drvvt)-based screening tests in all 5 proficiency testing surveys. Although mean clotting times varied, screening test results were relatively tightly distributed around the mean for each assay-method combination. As expected, the greatest screening test prolongation was observed for sample 2008-1, which contained the strong LAC. Conversely, the Table 1 Assay and Method Combinations No. of Results Evaluated Code Test Test Type Screen Mix Confirm 29 IL APTT SP APTT 24 13 0 88 IL HemosIL SynthASil APTT 7 8 0 21 Siemens/Dade Behring Actin FSL APTT 48 31 0 56 Stago/Roche PTT Automate APTT 34 36 0 57 Stago/Roche PTT LA APTT 79 51 0 43 Trinity/BioMerieux Platelin L APTT 24 12 0 1/2 American Diagnostica DVVtest/confirm drvvt 30 20 27 31/ 32 IL LAC screen/confirm drvvt 15 0 18 27/ 28 Life Diagnostics LA screen/confirm drvvt 9 0 11 80/ 81 Precision BioLogic LA check/sure drvvt 102 50 90 101/ 77 r 2 Diagnostics DRVVT screen/confirm drvvt 0 0 7 5/6 Siemens/Dade Behring LA1 screening/la2 confirmation drvvt 28 15 23 95 Precision Biologic Platelet Lysate PNP 0 0 39 59 Stago/Roche Staclot PNP PNP 0 0 15 58 Stago/Roche Staclot LA Integrated 0 0 93 APTT, activated partial thromboplastin time; drvvt, dilute Russell viper venom time; LAC, lupus anticoagulant; PNP, platelet neutralization procedure; PTT, partial thromboplastin time. Locations for companies are as follows: Instrumentation Laboratory (IL), Lexington, MA; Siemens/Dade Behring, Tarrytown, NY; Stago/Roche, Asnieres, France; Trinity/BioMerieux, Berkeley Heights, NJ; American Diagnostica, Stamford, CT; Life Diagnostics, Leiden, Netherlands; Precision BioLogic, Dartmouth, Canada; and r 2 Diagnostics, South Bend, IN. Evaluation of LAC testing performed in response to external proficiency challenges distributed by the North American Specialized Coagulation Laboratory Association in 2008 and 2009-2 required at least 3 reported results. Reagent sensitive to LAC. Reagent moderately sensitive to LAC. Table 2 Summary of Screening Test Results High-Titer LAC Medium-Titer LAC Low-Titer LAC Medium-Titer LAC Normal Plasma Pool Plasma Sample Plasma Pool Plasma Sample (Diluted) Plasma Pool Assay/ Result Result Result Result Result Method (seconds) No. (seconds) No. (seconds) No. (seconds) No. (seconds) No. APTT 21 47.9 ± 2.5 8 42.1 ± 3.9 9 30.7 ± 2.3 10 39.0 ± 3.5 11 28.1 ± 1.6 10 29 129.3 ± 25.0 3 56.5 ± 0.7 4 43.5 ± 3.5 6 49.3 ± 1.6 5 30.3 ± 1.6 6 57 104.5 ± 16.9 19 61.1 ± 4.5 17 51.6 ± 5.3 17 55.0 ± 4.1 15 39.8 ± 2.7 11 43 68.5 ± 6.5 6 49.2 ± 3.0 4 30.9 ± 1.2 6 45.7 ± 1.7 5 27.5 ± 1.1 3 56 64.0 ± 4.4 8 48.5 ± 2.2 7 36.8 ± 2.0 8 43.5 ± 2.5 7 32.3 ± 2.5 4 88 75.2 ± 3.5 4 49.8 ± 7.3 3 drvvt 1 90.3 ± 17.66 6 45.2 ± 7.6 7 45.9 ± 6.8 8 41.8 ± 5.1 6 37.6 ± 7.8 3 5 76.8 ± 2.9 5 68.6 ± 4.6 7 45.4 ± 1.8 6 54.7 ± 4.4 6 37.4 ± 1.2 4 27 56.6 ± 7.8 3 31 62.7 ± 5.2 5 44.1 ± 5.5 4 52.5 ± 4.2 3 34.8 ± 1.4 3 80 70.7 ± 5.2 22 60.2 ± 4.5 23 42.2 ± 9.5 22 51.7 ± 5.2 23 38.3 ± 2.8 12 APTT, activated partial thromboplastin time; drvvt, dilute Russell viper venom time; LAC, lupus anticoagulant. Screening test results were analyzed by assay and method. Assay/method combinations with 3 or more result submissions were included. Statistically significant differences (P <.05). Results of test samples (mean ± SD) were compared with those reported for local reference plasma using the unpaired Student t test. 766 Am J Clin Pathol 2010;134:764-773 766 DOI: 10.1309/AJCP4SPPLG5XVIXF

Coagulation and Transfusion Medicine / Original Article least prolongation was observed for sample 2009-2, normal plasma. Greater variability was noted in the ability of different assay and method combinations to identify intermediate- and low-titer LAC in samples 2008-2, 2008-3, and 2008-4. Although mean clotting times reported for samples containing LAC were significantly increased (P <.05), the extent of screening test prolongation varied by assay-method combination. A comparison of screening test performance for LAC detection is shown in Table 3. Although most assay and method combinations detected differences in LAC titer across the 5 proficiency testing challenges, the data suggest that APTT-based screening tests, particularly LAC-sensitive APTT methods, were more sensitive to intermediate- and low-titer LAC than drvvt-based methods. However, the false-positive rate was slightly higher. Indeed, all of the falsepositive results were observed with LAC-sensitive reagents. These results, however, need to be interpreted with caution because the overall numbers are low, with 1 or 2 false-negative or false-positive results per category. Results of mixing studies are summarized in Table 4. To compare the sensitivity of different assay and method combinations to LAC, we calculated the ICA 15 for all results that were accompanied by the clotting time of a normal plasma sample tested in like manner. ICA values exceeding 15 were considered suggestive of the presence of an LAC. 15 Significant variability in ICA scores can be appreciated between different assay and method combinations, particularly for samples containing weaker LAC. Based on self-reported interpretations of mixing studies, Table 3 Performance of Major LAC Screening Assays High-Titer LAC Medium-Titer LAC Low-Titer LAC Medium-Titer LAC Normal Assay Plasma Pool Plasma Sample Plasma Pool Plasma Sample (Diluted) Plasma Pool False-Negative (%) False-Negative (%) False-Negative (%) False-Negative (%) False-Positive (%) All 0 0 5.9 9.6 6.6 APTT (combined) 0 0 4.5 2.4 5.4 APTT (LAC sensitive) 0 0 3.0 3.2 7.4 APTT (LAC moderate sensitivity) 0 0 7.1 0 0 drvvt 0 0 7.5 17 7.8 APTT, activated partial thromboplastin time; drvvt, dilute Russell viper venom time; LAC, lupus anticoagulant. Performance of APTT and drvvt screening assays in LAC testing challenges was examined by determining false-positive and false-negative rates for each assay type based on participant response of normal (negative) or abnormal or borderline (positive). Table 4 Summary of Mixing Study Results High-Titer Intermediate-Titer Low-Titer LAC Intermediate-Titer LAC Normal LAC Plasma Pool LAC Plasma Sample Plasma Pool Plasma Sample (Diluted) Plasma Pool Assay/Method ICA No. ICA No. ICA No. ICA No. ICA No. APTT 21 15.6 ± 15.8 7 13.7 ± 12.2 10 8 ± 5.5 3 11.9 ± 4.4 8 5.85 ± 1.89 3 29 19 ± 2.5 4 24.5 ± 10.2 4 16.9 ± 3.2 5 57 40.0 ± 7.1 13 23.9 ± 13 8 23.3 ± 6.6 13 21.5 ± 7.1 13 7.2 ± 2.63 4 43 24.8 ± 4.0 5 23.2 ± 1.3 3 29.3 ± 17.5 4 56 31.1 ± 3.4 9 18.6 ± 1.9 7 14.7 ± 6.6 8 18.4 ± 5.8 8 3.31 ± 0.85 4 88 39.9 ± 3.5 4 19.3 ± 2.9 4 drvvt 1 35.7 ± 3.2 5 26.3 ± 16.6 4 10.4 ± 2.1 7 10.9 ± 2.4 4 5 38.7 ± 7.1 3 20.3 ± 1.8 4 10.1 ± 2.0 3 21.1 ± 5.0 5 80 66.8 ± 32.5 13 15.0 ± 3.6 14 5.2 ± 3.4 8 12.4 ± 1.9 11 3.69 ± 6.63 4 APTT, activated partial thromboplastin time; drvvt, dilute Russell viper venom time; LAC, lupus anticoagulant. Mixing study results for LAC evaluation were compared using the index of circulating anticoagulant (ICA). The ICA was calculated based on results submitted for sample screen, sample mix, and reference plasma mix, according to Rosner et al. 15 Results represent mean ± SD. Assay/method combinations with 3 or more reported results were included in the analysis. ICA values greater than 15; considered indicative of LAC. The negative calculated ICA reflects reported sample mix results (by 3 of 4 participants) that were shorter than results for the local normal plasma pool used in the mix. Am J Clin Pathol 2010;134:764-773 767 767 DOI: 10.1309/AJCP4SPPLG5XVIXF 767

Dembitzer et al / Lupus Anticoagulant Testing all participants identified the presence of LAC in sample 2008-1. However, when the calculated ICA was applied, an overall false-negative rate of 3.4% was noted. All false-negative results for this proficiency testing sample occurred with 1 LAC-sensitive reagent (method 21; Table 4). Detection of weak LAC in samples 2008-2, 2008-3, and 2008-4 varied for APTT- and drvvt-based methods using self-reported or calculated (ICA) interpretations. All results for sample 2009-2 were consistent with the absence of LAC. The comparison of mixing-study sensitivity for LAC detection is summarized in Table 5. Although variability in performance was observed across methods, when the ICA was used for mixing-study interpretation, APTT-based mixing studies appeared to be more sensitive to weak LAC than drvvt-based mixing studies. However, performance was similar when local interpretations were used for mixing method comparisons. Results of confirmatory testing are summarized in Table 6. Testing was performed by using a variety of commercial and home-made reagents and methods. Results for those assays and methods with greater than 3 data submissions are shown and represent automated testing exclusively. The integrated assay, method 58 (Stago/Roche, Staclot LA, Asnieres, France), was the most frequently used confirmatory assay, with 18 to 23 laboratories reporting results. All laboratories using this method correctly identified the strong lupus anticoagulant in sample 2008-1 and mostly identified weak LAC in sample 2008-3 but not in samples 2008-2 and 2008-4. It is interesting that 4 laboratories using this method incorrectly Table 5 Comparison of Mixing Studies for LAC Detection High-Titer LAC Intermediate-Titer Low-Titer LAC Intermediate-Titer LAC Normal Plasma Pool LAC Plasma Sample Plasma Pool Plasma Sample (Diluted) Plasma Pool False-Negative (%) False-Negative (%) False-Negative (%) False-Negative (%) False-Positive (%) Assay ICA Self ICA Self ICA Self ICA Self ICA Self All 3.4 0 19 7.6 39 5.6 32 32.8 0 0 APTT (combined) 5.2 0 23.5 12 22 3.1 28 32.6 0 0 APTT (LAC sensitive) 10 0 17.8 15 25 4.7 31 38.7 0 0 APTT (LAC moderate 0 0 9 6.6 36 0 25 20 0 0 sensitivity) drvvt 0 0 25 0 100 9.5 44 33 0 0 APTT, activated partial thromboplastin time; drvvt, dilute Russell viper venom time; ICA, index of circulating anticoagulant; LAC, lupus anticoagulant. Mixing studies performed using APTT and drvvt methods were compared for their ability to detect LAC by determining false-positive and false-negative rates for each assay type based on calculated ICA results or self-reported results for each participant. Table 6 Summary of Confirmatory Test Results High-Titer LAC Intermediate-Titer Low-Titer LAC Intermediate-Titer LAC Normal Plasma Pool LAC Plasma Sample Plasma Pool Plasma Sample (Diluted) Plasma Pool Assay/Method + B + B + B + B + B Integrated 58 18 0 0 3 1 19 1 19 1 0 2 16 0 4 9 drvvt 2 4 0 0 0 2 4 0 6 2 0 0 5 0 0 4 6 4 0 0 6 0 0 1 4 1 0 4 0 0 0 3 28 2 0 0 1 1 0 0 4 0 0 1 1 0 0 1 32 3 0 0 3 0 1 0 2 2 0 1 3 0 0 3 77 2 0 0 0 0 1 1 0 0 1 0 1 0 0 1 81 20 0 1 16 2 6 3 8 5 2 5 13 0 1 8 PNP 59 4 0 0 0 0 3 0 2 0 1 0 3 0 1 1 95 7 0 0 1 0 9 2 2 3 1 0 7 0 0 7 drvvt, dilute Russell viper venom time; LAC, lupus anticoagulant; PNP, platelet neutralization procedure. Participant interpretations of confirmatory test results are listed according to assay type and method. Interpretations were reported as positive (+), borderline (B), or negative ( ). Data are given as number. 768 Am J Clin Pathol 2010;134:764-773 768 DOI: 10.1309/AJCP4SPPLG5XVIXF

Coagulation and Transfusion Medicine / Original Article identified a borderline LAC in sample 2009-2, the normal plasma sample. drvvt confirmatory testing also uniformly identified the strong LAC in sample 2008-1, and laboratories reported predominantly borderline and negative results for samples 2008-2, 2008-3, and 2008-4. Only 1 false-positive (borderline LAC) was reported for sample 2009-2. Assays based on the platelet neutralization procedure (PNP) identified the strong LAC in sample 2009-1 but showed overall higher false-negative rates for detecting weak LAC compared with integrated APTT or drvvt methods. Moreover, 1 borderline positive LAC result was reported for the normal sample using a PNP-based method. A comparison of confirmatory assay performance is provided in Table 7. NASCOLA/ECAT proficiency testing challenges for LAC also require laboratories to render an overall assessment/ final diagnosis of the presence or absence of LAC based on all testing performed. Five interpretive choices are offered: clearly positive, positive, probably positive, borderline, and negative. For purposes of this study, reports of clearly positive, positive, and probably positive were grouped as positive. A summary of final interpretations is shown in Table 8. As expected, interpretations reflected the strength of the LAC present in the individual proficiency testing samples. All laboratories identified the strong LAC in sample 2008-1, but identification of weaker lupus anticoagulants in samples 2008-2, 2008-3, and 2008-4 was more variable. Overall, false-negative rates for samples 2008-1, 2008-2, 2008-3, and 2008-4 were 0%, 28%, 25%, and 24%, respectively. Borderline results were considered indicative of LAC in these samples. To evaluate the false-positive rates of LAC detection by NASCOLA laboratories, a normal sample was circulated for testing. Although no positive results were reported by major Table 7 Performance of Confirmatory Assays for LAC Identification High-Titer LAC Intermediate-Titer Low-Titer LAC Intermediate-Titer LAC Normal Plasma Pool LAC Plasma Sample Plasma Pool Plasma Sample (Diluted) Plasma Pool Assay False-Negative (%) False-Negative (%) False-Negative (%) False-Negative (%) False-Positive (%) Integrated 0 82 5 89 31 drvvt 3 29 26 62 5 PNP 0 92 33 83 11 drvvt, dilute Russell viper venom time; LAC, lupus anticoagulant; PNP, platelet neutralization procedure. The false-negative and false-positive rates for assay types were calculated based on confirmatory test results shown in Table 6. Table 8 Summary of Final LAC Testing Interpretation and Laboratory Practices High-Titer LAC Intermediate-Titer Low-Titer LAC Intermediate-Titer LAC Normal Plasma Pool LAC Plasma Sample Plasma Pool Plasma Sample (Diluted) Plasma Pool Final interpretation (No.) 49 50 53 50 46 Positive 49 27 35 34 2 Negative 0 14 13 12 41 Borderline 0 9 5 4 3 No. (%) of noncompliance events 4 (8) 10 (20) 20 (38) 11 (22) 10 (22) No mix (No.) 4 5 13 6 1 No mix/confirm (No.) 0 2 5 1 1 Insufficient testing (No.) 0 3 2 4 8 Misdiagnosis (%) All laboratories 0 28 25 24 11 Compliant laboratories 0 24 16 19 3 Noncompliant laboratories 0 30 39 40 3 No. of tests performed (mean ± SD) Positive result 5.9 ± 1.5 6.1 ± 1.8 6.3 ± 1.8 6.3 ± 1.7 7.1 Negative result 6.3 ± 1.3 4.8 ± 1.3 5.8 ± 1.1 3.85 ± 2.04 Borderline result 7.1 ± 1.6 5.3 ± 1.0 5.7 ± 0.6 3.67 ± 1.15 LAC, lupus anticoagulant. Includes the following interpretations: clearly positive, positive, and probably positive. False-positive or false-negative results. Am J Clin Pathol 2010;134:764-773 769 769 DOI: 10.1309/AJCP4SPPLG5XVIXF 769

Dembitzer et al / Lupus Anticoagulant Testing assay-method combinations for this normal plasma sample (2009-2), the overall false-positive rate for this exercise was approximately 11% if borderline results were included as evidence of LAC. It is interesting that the greatest number of false-positive results (borderline) was reported by participants using integrated APTT method 58. The lowest false-positive rate occurred with drvvt-based methods. In addition, incidences of misinterpretation of laboratory test results were noted for samples containing weak LAC (2008-2, 2008-3, and 2008-4). A minority of laboratories (1-2 in each proficiency testing challenge) reported final interpretations that were inconsistent with confirmatory test results: for example, a positive final interpretation was reported with negative confirmatory test results and vice versa. Consistent with the presence of weak LAC in samples 2008-3 and 2008-4, 3 and 2 laboratories, respectively, interpreted either positive or negative confirmatory test results as borderline in the final diagnosis. A total of 25 laboratories incorrectly assessed the presence or absence of LAC in proficiency testing challenges 2008-2, 2008-3, 2008-4, and 2009-2. None of these laboratories reported false-negative results for sample 2008-1 containing the strong LAC. Thirteen laboratories reported 1 incorrect LAC diagnosis. Five participating laboratories reported false-negative results in 3 consecutive testing periods for samples containing weak LACs. Seven laboratories reported false-negative results across 2 consecutive testing periods. The 5 participants who reported false-positive/borderline results for sample 2009-2 correctly identified LAC in all 2008 challenges. It is interesting that no difference was noted in the extent of testing, ie, number of tests performed by laboratories reporting correct and incorrect LAC identifications (Table 8). The overall reduction in tests performed for sample 2009-2 reflects a decrease in mixing studies performed for this sample with overwhelmingly normal screening test results. Compliance with ISTH guidelines for evaluating LAC was variable among the testing laboratories and for the 5 proficiency testing periods (Table 8). Noncompliance was lowest (8%) for proficiency testing challenge 2008-1 and highest (38%) for proficiency testing challenge 2008-3. In the majority of cases, laboratories that were not compliant failed to perform mixing studies. A total of 27 individual laboratories failed to comply with testing guidelines in at least 1 testing period. Fifteen laboratories failed to comply with testing guidelines more than once, with 1 laboratory being consistently noncompliant in all 5 testing challenges. Based on reported results, 3 laboratories were deemed noncompliant in 4 testing periods and 5 laboratories in 2 and 3 test periods, each. Paradoxically, an analysis of testing outcomes/performance by compliance indicated that overall, compliant laboratories had substantial false-negative rates (Table 8). Discussion The laboratory diagnosis of LAC is challenging. 16,17 This is due, in large part, to the heterogeneity of LACs reacting with poorly characterized epitopes on proteins associated with negatively charged phospholipids. Currently available laboratory assays and methods demonstrate substantial differences in their ability to detect LACs, 3,17,18 and no single LAC test is capable of detecting all LACs. 3,5,19 Indeed, surveys performed predominantly in Europe and Australia during the last 10 years have shown variable sensitivity and specificity of LAC tests and suboptimal laboratory performance. 9,17,20-22 The present study examined LAC testing patterns and performance by specialized coagulation laboratories in the United States and Canada participating in NASCOLA/ECAT external proficiency testing challenges. North American specialized coagulation laboratories relied predominantly on APTT- and drvvt-based assays to screen for LAC. Screening tests are dependent on phospholipids, and their sensitivity to LAC varies according to the composition of reagents, the class and concentration of phospholipids, and phospholipid conformation. Although a variety of methods were used, Siemens/Dade Behring (Tarrytown, NY) Actin FSL (APTT, LAC-sensitive reagent) and Stago/Roche (drvvt) were the most popular. Consistent with previous reports, 3,17,18 considerable variability in sensitivity was noted between methods. Strong LACs were usually identified, but a significant drop off occurred in the presence of intermediate- and low-titer LACs. Although proficiency testing samples were characterized as containing strong and weak LACs, it should be noted that it is unclear whether high-titer LACs are stronger risk factors for thrombotic complications than low-titer LACs. Moreover, the laboratory tests themselves are not quantitative, and there are no criteria to define weak positives and strong positives. 23 It has been suggested that drvvt-based testing is more sensitive and specific for detecting LAC, especially in patients at high risk for developing thrombosis. In recent years, the sensitivity of APTT-based testing has improved with modifications to the assay phospholipids. However, its specificity is still debated. In the present study, drvvt- and APTT-based screening tests showed differences in sensitivity that appeared to be sample/lac related. Based on the results of 1 proficiency challenge (2009-2), drvvt testing appeared to be more specific for LAC (fewer false-positive results) compared with APTT-based testing. As mentioned previously, however, these data need to be interpreted with some caution because the number of data points for individual methods in this study was relatively small and the number of false-positive and false-negative results exceedingly low. 1-3 Mixing studies are recommended as part of LAC testing by ISTH guidelines. 13,14 Mixing studies are based on the rationale that a mixture of equal amounts of patient plasma and a 770 Am J Clin Pathol 2010;134:764-773 770 DOI: 10.1309/AJCP4SPPLG5XVIXF

Coagulation and Transfusion Medicine / Original Article plasma pool derived from healthy people will considerably shorten or correct the prolonged patient coagulation time if it is due to a deficiency of 1 or more coagulation factors. Conversely, the mixture does not correct the prolonged clotting time when it is due to the presence of an inhibitor. Although simple in principle, the mixing study has considerable drawbacks, including the quality of normal plasma used. Revised 2009 ISTH guidelines state that the normal pool must contain 100% of all clotting factors. Since the source of normal plasma used for mixing studies was not captured on NASCOLA result reports, we were unable to determine whether normal plasma had been assayed for individual coagulation factors. In addition, problems exist with regard to interpretation of mixing studies. 5 The revised 2009 ISTH guidelines 14 recommend using a cutoff for correction that is beyond the 99th percentile of the distribution or calculating an ICA. The present study was unable to determine the local criteria used by North American laboratories to interpret results of mixing studies. Because the ICA is generally considered the most robust, 5 we calculated the ICA to compare results of mixing studies by assay and method. It is interesting that interpretation of mixing study results by individual laboratories using local criteria did not correlate well with results of ICA calculations, particularly for samples 2008-2 and 2008-3. In general, false-negative rates were lower for self-reported mixing study interpretations than for interpretations using the calculated ICA, suggesting that locally established cutoffs improve performance. Considerable variation was noted in the ability of APTTor drvvt-based mixing studies to detect weak LACs in samples 2008-2, 2008-3, and 2008-4. Moreover, no single assay consistently performed better than any other. Indeed, the usefulness of mixing studies has been challenged, 24 particularly because weak LACs may be missed owing to dilutional effects when performing mixing studies, leading to false-negative results. 3,4 It has been suggested that mixing tests be applied to APTT and drvvt screening and confirmatory tests for LAC to ensure correct interpretation. 3 Despite ISTH recommendations, a number of North American laboratories failed to perform mixing studies. Indeed, the failure to perform mixing studies was the reason most often identified as to why a laboratory was considered noncompliant with regard to following LAC testing recommendations. ISTH guidelines further recommend the use of confirmatory tests for the diagnosis of LAC. The rationale for using confirmatory tests is that increasing the concentration of phospholipids in the test system will neutralize the effect of LAC and shorten the prolonged coagulation time if it is due to the presence of LAC. North American laboratories used a combination of drvvt- and APTT-based confirmatory methods. The use of integrated systems has gained in popularity, 5,16 with many North American laboratories using the APTT-based Staclot LA procedure (Stago/Roche), which performs LAC testing in the presence of a mixture of patient and normal plasma. The intrinsic heterogeneity of these test systems is borne out by the significantly different results that were observed when the systems were compared, here and in the literature. 16 A significant number of laboratories also performed a PNP. In the four 2008 proficiency testing challenges with LAC-positive samples, PNP methods appeared to be the least sensitive. Indeed, 1 of the 2 PNP methods used by NASCOLA laboratories (method 59) is no longer available commercially. Misclassification of LAC (false-positive and false-negative results) by North American laboratories depended on the strength of the LAC, consistent with results reported for laboratories in the United Kingdom and Italy. 1,9 Whereas all laboratories correctly identified the LAC in sample 2008-1 containing the strong LAC, a misdiagnosis rate of approximately 25% was noted for samples with weak LAC. Combining LAC testing using the integrated assay (method 58) with a drvvt (method 81) may enhance sensitivity to some (samples 2008-2 and 2008-3) but not all (2008-4) weak LACs. While it has been suggested that performing more than 2 screening tests would yield too many false-positive test results, 25 the present study was unable to evaluate this because most laboratories performed 2 screening tests in accordance with ISTH recommendations, and only a single LAC-negative sample was available for analysis. It is interesting that of the laboratories that repeatedly rendered an incorrect LAC diagnosis for samples 2008-2, 2008-3, 2008-4, and/or 2009-2, only 2 showed evidence of changing methods. Because laboratory codes are confidential, we were unable to obtain information regarding institutional size and affiliation (eg, private laboratory, academic medical center) of laboratories that rendered correct and incorrect diagnoses. The overall noncompliance rate for North American laboratories in this study ranged from 8% to 38% across proficiency testing periods. Moreover, laboratories that were noncompliant repeatedly remained faithful to their testing practices. Thus, 5 laboratories were noncompliant during 3 testing periods, 2 laboratories were noncompliant in 4 proficiency testing challenges, and 1 laboratory was noncompliant across all 5 testing periods. Laboratory compliance with revised 2009 LAC testing guidelines 14 was difficult to assess because testing was performed before their publication. However, the majority of North American laboratories did not perform testing that is no longer recommended, including the dilute prothrombin time, ecarin and textarin times, and the kaolin clotting time. Although the use of frozen and thawed platelets in the confirmatory test Am J Clin Pathol 2010;134:764-773 771 771 DOI: 10.1309/AJCP4SPPLG5XVIXF 771

Dembitzer et al / Lupus Anticoagulant Testing also is no longer recommended, PNP methods were the third most widely used testing platform (n = 11-17) for LAC confirmation. The present study provides an overview of the state-ofthe-art of LAC testing in North American laboratories participating in NASCOLA proficiency testing surveys. Consistent with the international experience, detection of weak LACs is problematic, and strict adherence to ISTH guidelines could be improved. As suggested previously, more specific LAC test procedures are needed, but their development requires a better understanding of the mechanism(s) associated with clinical events. Although not addressed in this study, the effect of preanalytic variables, such as factor inhibitors or anticoagulants, adds complexity to testing and must be carefully evaluated for each method-assay combination. Additional studies are needed to evaluate these effects and to formulate more specific guidelines for test interpretation under these conditions. Alternatively or in addition, recommendations for when testing should and should not be performed would be helpful for physicians and laboratories. The following recommendations are proposed for improving LAC testing. Sensitivity and specificity of testing may be improved by establishing laboratory-specific reference ranges and cutoffs for screening and confirmatory tests, as suggested in the 2009 ISTH guidelines. 14 Moreover, compliance with these same guidelines to perform combination testing with drvvt- and sensitive APTT-based methods may improve sensitivity to weak LACs. Observations from our study show a heterogeneous pattern for detecting weak LACs with various reagents, thereby supporting, at least for the interim, the use of a panel of reagents. However, a well controlled study wherein a number of patient samples containing weak LACs are challenged with a variety of reagents from various manufacturers may provide some insight as to which reagents are more robust in identifying weak LACs. Additionally, the standardization of mixing study interpretations and the role of mixing studies as a screening test for LACs should be assessed in studies where laboratory and clinical data can be considered together. Finally, an overall interpretation of all LAC testing, not only the interpretation of individual tests, together with a patient s clinical information is required to make an appropriate diagnosis. From the 1 Department of Pathology, Mount Sinai School of Medicine, New York, NY; 2 ECAT Liaison to NASCOLA, Islamorada, FL; and 3 ECAT Foundation, Leiden, the Netherlands. Supported by funds from the Division of Translational and Applied Laboratory Medicine, Department of Pathology, Mount Sinai Medical Center, New York. Address reprint requests to Dr Peerschke: Center for Clinical Laboratories, 1425 Madison Ave, Room 8-02 A, New York, NY 10029. Acknowledgment: We thank Samir Zaman for assistance with data evaluation. References 1. Jennings I, Kitchen S, Woods TAL, et al; for the UK National External Quality Assessment Scheme for Blood Coagulation. Potentially clinically important inaccuracies in testing for the lupus anticoagulant: an analysis of results from three surveys of the UK National External Quality Assessment Scheme (NEQAS) for Blood Coagulation. Thromb Haemost. 1997;77:934-937. 2. Giannakipoulos B, Passam F, Ioannou U, et al. How we diagnose the antiphospholipid syndrome. Blood. 2009;113:985-994. 3. Devreese K, Hoylaerts MF. Laboratory diagnosis of the antiphospholipid syndrome: a plethora of obstacles to overcome. Eur J Haematol. 2009;83:1-16. 4. Moffat KA, Ledford-Kraemer MR, Plumhoff EA, et al. Are laboratories following published recommendations for lupus anticoagulant testing? an international evaluation of practices. Thromb Haemost. 2009;101:178-184. 5. Tripodi A. Laboratory testing for lupus anticoagulants: a review of issues affecting results. Clin Chem. 2007;53:1629-1635. 6. Wong RCW; for the Australasian acl Working Party. Consensus guidelines for anticardiolipin antibody testing. Thromb Res. 2004;114:559-571. 7. Tripodi A, Chantarangkul V, Clerici M, et al. Laboratory diagnosis of lupus anticoagulants for patients on oral anticoagulant treatment: performance of dilute Russell viper venom test and silica clotting time in comparison with Staclot LA. Thromb Haemost. 2002;88:583-586. 8. Tripoldi A, Biasiolo A, Chantarangkul V, et al. Lupus anticoagulant (LA) testing: performance of clinical laboratories assessed by a national survey using lyophilized affinity-purified immunoglobulin with LA activity. Clin Chem. 2003;49:1608-1614. 9. Pengo V, Biasiolo A, Gresele P, et al; for the Participating Centres of Italian Federation of Thrombosis Centres (FCS). Survey of lupus anticoagulant diagnosis by central evaluation of positive plasma samples. J Thromb Haemost. 2007;5:925-930. 10. Favaloro EJ, Wong RCW. Laboratory testing and identification of antiphospholipid antibodies and the antiphospholipid syndrome: a potpourri of problems, a compilation of possible solutions. Semin Thromb Hemost. 2008;34:389-410. 11. Jennings I, Greaves M, Mackie IJ, et al; for the UK National External Quality Assessment Scheme for Blood Coagulation. Lupus anticoagulant testing: improvements in performance in a UK NEQAS proficiency testing exercise after dissemination of national guidelines on laboratory methods. Br J Haematol. 2002;119:364-369. 12. Brandt JT, Triplett DA, Rock WA, et al. Effect of lupus anticoagulants on the activated partial thromboplastin time. Arch Pathol Lab Med. 1991;115:109-114. 13. Brandt JT, Triplett DA, Alving B, et al. Criteria for the diagnosis of lupus anticoagulants: an update, on behalf of the subcommittee on Lupus anticoagulant/antiphospholipid Antibody of the Scientific and Standardisation Committee of the ISTH. Thromb Haemost. 1995;74:1185-1190. 772 Am J Clin Pathol 2010;134:764-773 772 DOI: 10.1309/AJCP4SPPLG5XVIXF

Coagulation and Transfusion Medicine / Original Article 14. Pengo V, Tripodi A, Reber G, et al. Update of the guidelines for lupus anticoagulant detection. J Thromb Haemost. 2009;7:1737-1740. 15. Rosner E, Pauzner R, Lusky A, et al. Detection and quantitative evaluation of lupus circulating anticoagulant activity. Thromb Haemost. 1987;57:144-147. 16. Triplett DA. Use of the dilute Russell viper venom time (drvvt): its importance and pitfalls. J Autoimmun. 2000;15:173-178. 17. Arnout J, Meijer P, Vermylen J. Lupus anticoagulant testing in Europe: an analysis of results from the First European Concerted Action on Thrombophilia (ECAT) Survey using plasmas spiked with monoclonal antibodies against human beta2-glycoprotein I. Thromb Haemost. 1999;81:929-934. 18. Wong RCW, Favaloro EJ. A consensus approach to the formulation of guidelines for laboratory testing and reporting of antiphospholipid antibody assays. Semin Thromb Hemost. 2008;34:361-372. 19. Triplett DA. Antiphospholipid antibodies. Arch Pathol Lab Med. 2002;126:1424-1429. 20. Roussi J, Roisin JP, Goguel A. Lupus anticoagulants: first French interlaboratory Etalonorme survey. Am J Clin Pathol. 1996;105:788-793. 21. Favaloro EJ, Bonar R, Sioufi J, et al. Multi-laboratory testing of thrombophilia: current and past practice in Australasia as assessed through the Royal College of Pathologists of the Australasia Quality Assurance Programs for Hematology. Semin Thromb Hemost. 2005;31:49-58. 22. Favaloro EJ, Bonar R, Duncan E, et al; on behalf of the RCPA QAP in Haematology Haemostasis Committee. Identification of factor inhibitor by diagnostic haemostasis laboratories: a large multicenter evaluation. Thromb Haemost. 2006;96:73-78. 23. Teruya J, West AG, Suell MN. Lupus anticoagulant assays: questions answered and to be answered. Arch Pathol Lab Med. 2007;131:885-889. 24. Ledford-Kraemer MR. Laboratory testing for lupus anticoagulants: preexamination variables, mixing studies, and diagnostic criteria. Semin Thromb Hemost. 2008;34:380-388. 25. Tripodi A. Laboratory testing for lupus anticoagulants: diagnostic criteria and use of screening, mixing, and confirmatory studies. Semin Thromb Hemost. 2008;34:373-378. Am J Clin Pathol 2010;134:764-773 773 773 DOI: 10.1309/AJCP4SPPLG5XVIXF 773