Haemophilia (2014), 20 (Suppl. 4), 94–98
DOI: 10.1111/hae.12408
REVIEW ARTICLE
Laboratory testing for factor inhibitors E . J . F A V A L O R O , * B . V E R B R U G G E N † ‡ and C . H . M I L L E R § *Diagnostic Haemostasis, Haematology Department, Institute of Clinical Pathology and Medical Research (ICPMR), Pathology West, Westmead Hospital, Westmead, NSW, Australia; †ECAT Foundation, Leiden, The Netherlands; ‡Laboratory of Clinical Chemistry and Haematology, Jeroen Bosch Hospital, sHertogenbosch, The Netherlands; and §Division of Blood Disorders, National Center on Birth Defects and Developmental Disabilities, Centers for Disease Control and Prevention, Atlanta, GA, USA
Summary. Inhibitor assays are performed when patients present with unexplained prolonged routine coagulation test times and unexpected and/or unusual bleeding (potential for acquired haemophilia) as well as being a part of normal congenital haemophilia management and monitoring, particularly when bleeding occurs on therapy, or when increments in factor levels post-factor replacement remain lower than expected. In this article, we will describe the assays used, as well as their development, pitfalls in testing such as inter-laboratory variability and false negative/ positive results, as well as some strategies for overcoming these pitfalls and potential alternative test
approaches. The inter-laboratory coefficient of variation often approaches (and sometimes exceeds) 50%, as evidenced by various external quality assessment groups, and this variability has not improved over recent years. Additional important considerations include appropriate interpretation of test results, repeat testing for confirmation, and assessment of recovery as part of the diagnostic process.
Introduction – Bethesda to Nijmegen and beyond (EJF, BV, CHM)
tion in the Nijmegen assay (NA), by buffering the NPP (BNPP) and replacing the imidazole buffer with inhibitor-free FVIII-deficient plasma as reference sample [4]. Additional recommendations, including heating of test and reference plasma to remove residual FVIII [5], the appropriate use of FVIII-deficient reference plasma [6], the use of 4M imidazole solution instead of solid imidazole for buffering the incubation mixture and the determination of residual FVIII activity with chromogenic substrates, continue to improve the sensitivity and specificity of the assay.
The Bethesda assay (BA) for factor(F)VIII inhibitors, described in 1975 by Kasper and colleagues [1], was the first method yielding an acceptable degree of standardization, using normal pooled plasma (NPP) as FVIII source and imidazole buffer as reference sample. The BA replaced the Oxford assay [2], which lacked reliability as it used FVIII concentrate (cryoprecipitate) as FVIII source. Since the late 1980s, inhibitor assays were performed with increasing frequency, largely driven by multicentre studies of new purified and recombinant FVIII products. The BA soon appeared to be rather non-specific, yielding many false positive results [3]. Its sensitivity and specificity were improved by modificaCorrespondence: E. J. Favaloro, Diagnostic Haemostasis, Haematology Department, Institute of Clinical Pathology and Medical Research (ICPMR), Pathology West, Westmead Hospital, Westmead, NSW, Australia. Tel.: +612 9845 5161 (Office/Messages), +612 9845 6618 (Laboratory); fax: +612 9689 2331; e-mail: [email protected] Accepted 6 February 2014 94
Keywords: assay variation, diagnostic accuracy, fluorescence immunoassay, factor inhibitors, haemophilia, low titre inhibitors
Pitfalls and recommendations on inhibitor testing – lessons from External Quality Assessment (EQA) (EJF, BV) Nevertheless, inter-laboratory coefficients of variation (CV) approaching (and sometimes exceeding) 50% are evidenced by various EQA groups, inclusive of ECAT (External quality Control of Assays and Tests) Foundation, NEQAS (UK National External Quality Assurance Scheme) and RCPA (Royal College of Pathologists of Australasia) (Fig. 1) [7–10]. This variability has not improved over the years. Reasons for such high variability largely derive from assay © 2014 John Wiley & Sons Ltd
INHIBITOR TESTING (a)
95
(b)
Fig. 1. Summary of recent EQA data highlighting inter-laboratory variation for factor inhibitors. (A) Data for FVIII inhibitor testing using composite data for RCPAQAP showing coefficient of variation (CV, %; y-axis) plotted against median participant reported inhibitor level (in Bethesda units [BU]/mL; x-axis) for all samples distributed over the past 6 years, and shown separately for Bethesda or Nijmegen methods. The average and 95% confidence intervals of the mean of all data are shown on the right of the figure. (B) Recent data for ECAT for FIX inhibitor testing showing the participant reported inhibitor level (y-axis; BU/mL) for two samples distributed in 2013 (identified on x-axis). Figure redrawn from reference 10.
variability, which can occur at any stage within the assay, including choice and source of laboratory reagents, buffering (yes, no, how), NPP, etc., such that no two laboratories participating in the RCPA EQA programme could be found to be using exactly the same procedure in one analysis [8]. Although NA tend to provide lower variability than BA [7–10], how laboratories perform NA varies almost as much as how laboratories perform BA; thus, variability still remains high. The percentage of falsely positive and falsely negative results in FVIII-inhibitor-negative samples is also unacceptably high (up to 32% for false positives; up to 5% for false negatives) [7–10]. To improve reproducibility, ECAT performed a quality improvement cycle including these steps: (i) an external survey among 51 laboratories that participated on a regular basis in the ECAT FVIII-inhibitor programme; (ii) selection from these of 15 representative laboratories for a centralized workshop; (iii) during the workshop, a zero point measurement using participants’ own methods and reagents and (iv) separate measurements with a fully standardized and universal method (NA) and reagents; (v) an external survey among 13 workshop participants 3 months after the initial workshop and (vi) an external survey among 22 of the original 51 laboratories with standardized methods (BNPP with FVIII activity ranging from 0.95 to 1.05 IU mL 1, FVIII-deficient plasma as reference sample, standardized sample dilution with FVIII-deficient plasma). In each step of the cycle, an identical set of seven samples was used (one negative and six positive). The means and CVs of results using the six inhibitorpositive samples at the various steps (Table 1) clearly show that very low inter-laboratory variations can be achieved using a centralized setting with uniform methods and reagents (step 4), and acceptable inter© 2014 John Wiley & Sons Ltd
laboratory variations are possible in EQA surveys by significant standardization of the methods (step 6). The inhibitor activity of the negative sample was also below the cut-off value in all participants bar one in steps 4 and 6. Additional pitfalls as well as strategies for improvements in this area of testing are extensively covered in a recent review [10]. In brief, pitfalls occur at any stage of the diagnostic process, including pre-analytical issues (doctors test request, sample collection), analytical (methodology as detailed above), and post-analytical (laboratory interpretation and reporting, doctors’ interpretation and action). Main strategies comprise: (a) pre-analytical: (i) check test orders for accuracy and relevance (to ensure performance of the correct tests); (ii) check and be aware of blood sampling issues (correct anticoagulant, proper fill, etc.; this will help avoid inaccuracies, and false positive/negative test findings); (b) analytical: (i) be aware of assay limitations such as low level inhibitor sensitivities, (ii) use the best methodologies and implement proven assay improvements (helps ensure accuracy), (iii) perform appropriate internal quality control and participate in EQA (helps ensure accuracy), (iv) repeat tests when necessary (helps ensure accuracy and avoid false negatives/positives); (c) post-analytical: (i) request repeat testing using fresh samples for confirmation when test results do not match expectations; (ii) provide consultative services with clinicians to assist in test interpretation.
Further enhancing sensitivity and specificity of inhibitor testing (BV) It may be concluded that implementation of more standardized methods will lead to better specificity, as Haemophilia (2014), 20 (Suppl. 4), 94--98
96
E. J. FAVALORO et al.
Table 1. Recent findings from an ECAT study.
Sample number (nominal inhibitor activity) 1 (1.6 BU mL 1) 2 (0.8 BU mL 1) 3 (1.4 BU mL 1) 4 (0.7 BU mL 1) 5 (1.9 BU mL 1) 6 (15.4 BU mL 1) Mean CV
Steps 1 + 2
Steps 3 + 4
Pre-workshop survey (2009)
Workshop results (2009)
Step 5
Step 6
Post-workshop survey (2010)
Standardized final survey (2012)
51 Laboratories
15 laboratories selected for the workshop
Zero measurement
Standardized measurement
13 Laboratories
22 Laboratories
2.32 (36%) 0.79 (49%) 0.9 (41%) 0.44 (70%) 1.74 (36%) 11.0 (36%) 45%
2.69 (43%) 1.02 (31%) 1.16 (39%) 0.59 (69%) 1.74 (37%) 11.5 (44%) 44%
2.97 (39%) 1.33 (69%) 1.17 (30%) 0.61 (45%) 2.34 (41%) 14.9 (41%) 44%
1.93 (8%) 0.94 (5%) 1.16 (6%) 0.50 (13%) 2.22 (12%) 14.6 (6%) 8%
2.9 (41%) 1.1 (88%) 1.1 (31%) 0.6 (61%) 1.9 (31%) 12.0 (36%) 48%
2.7 (31%) 0.7 (17%) 1.0 (23%) 0.5 (30%) 1.8 (22%) 12.4 (27%) 25%
Mean inhibitor activity in BU mL 1; inter-laboratory coefficient of variation (CV) given in brackets.
evidenced by the above ECAT study. Besides problems with reproducibility and specificity, both BA and NA lack sensitivity for low inhibitor activities. The internationally agreed detection limit for both tests is 0.6 Bethesda units (BU), although it may be lower for NA because of improved specificity. Nevertheless, both assays may miss inhibitors with low activity. From personal experience, it was hypothesized that these lowtitre (‘undetectable’) inhibitors might be clinically significant and present in patients in the late Immune Tolerance Induction (ITI) phase, rendering replacement therapy less effective and leading to bleeding complications and increased need of FVIII concentrates in these patients. Therefore, a low-titre FVIII inhibitor assay (LTA) was recently developed and described with a lower limit of detection of 0.03 BU [11]. The principle of the LTA is identical to NA except for the use of concentrated plasma instead of native plasma, an alternative ratio of concentrated plasma/BNPP in the test mixture of 3:1, and the use of chromogenic substrates for assay of residual FVIII. Assay results are expressed in BU by correcting the analysis data for the concentration factor of the plasma and the alternative ratio. Using LTA, low-titre inhibitors were demonstrated as still present in the early post-ITI phase in haemophiliacs treated for FVIII inhibitors despite negative findings with NA or BA. These low-titre inhibitors decrease the half-life and the recovery of infused FVIII products [11]. The clinical significance of LTA was further evaluated in a satellite study of the International ITI (I-ITI) study [12], in which inhibitor-positive patients were treated with low (50 IU kg 1 thrice weekly) or high (200 IU kg 1 per day) dose regimens of recombinant FVIII concentrates until the NA or BA became negative (
DOI: 10.1111/hae.12408
REVIEW ARTICLE
Laboratory testing for factor inhibitors E . J . F A V A L O R O , * B . V E R B R U G G E N † ‡ and C . H . M I L L E R § *Diagnostic Haemostasis, Haematology Department, Institute of Clinical Pathology and Medical Research (ICPMR), Pathology West, Westmead Hospital, Westmead, NSW, Australia; †ECAT Foundation, Leiden, The Netherlands; ‡Laboratory of Clinical Chemistry and Haematology, Jeroen Bosch Hospital, sHertogenbosch, The Netherlands; and §Division of Blood Disorders, National Center on Birth Defects and Developmental Disabilities, Centers for Disease Control and Prevention, Atlanta, GA, USA
Summary. Inhibitor assays are performed when patients present with unexplained prolonged routine coagulation test times and unexpected and/or unusual bleeding (potential for acquired haemophilia) as well as being a part of normal congenital haemophilia management and monitoring, particularly when bleeding occurs on therapy, or when increments in factor levels post-factor replacement remain lower than expected. In this article, we will describe the assays used, as well as their development, pitfalls in testing such as inter-laboratory variability and false negative/ positive results, as well as some strategies for overcoming these pitfalls and potential alternative test
approaches. The inter-laboratory coefficient of variation often approaches (and sometimes exceeds) 50%, as evidenced by various external quality assessment groups, and this variability has not improved over recent years. Additional important considerations include appropriate interpretation of test results, repeat testing for confirmation, and assessment of recovery as part of the diagnostic process.
Introduction – Bethesda to Nijmegen and beyond (EJF, BV, CHM)
tion in the Nijmegen assay (NA), by buffering the NPP (BNPP) and replacing the imidazole buffer with inhibitor-free FVIII-deficient plasma as reference sample [4]. Additional recommendations, including heating of test and reference plasma to remove residual FVIII [5], the appropriate use of FVIII-deficient reference plasma [6], the use of 4M imidazole solution instead of solid imidazole for buffering the incubation mixture and the determination of residual FVIII activity with chromogenic substrates, continue to improve the sensitivity and specificity of the assay.
The Bethesda assay (BA) for factor(F)VIII inhibitors, described in 1975 by Kasper and colleagues [1], was the first method yielding an acceptable degree of standardization, using normal pooled plasma (NPP) as FVIII source and imidazole buffer as reference sample. The BA replaced the Oxford assay [2], which lacked reliability as it used FVIII concentrate (cryoprecipitate) as FVIII source. Since the late 1980s, inhibitor assays were performed with increasing frequency, largely driven by multicentre studies of new purified and recombinant FVIII products. The BA soon appeared to be rather non-specific, yielding many false positive results [3]. Its sensitivity and specificity were improved by modificaCorrespondence: E. J. Favaloro, Diagnostic Haemostasis, Haematology Department, Institute of Clinical Pathology and Medical Research (ICPMR), Pathology West, Westmead Hospital, Westmead, NSW, Australia. Tel.: +612 9845 5161 (Office/Messages), +612 9845 6618 (Laboratory); fax: +612 9689 2331; e-mail: [email protected] Accepted 6 February 2014 94
Keywords: assay variation, diagnostic accuracy, fluorescence immunoassay, factor inhibitors, haemophilia, low titre inhibitors
Pitfalls and recommendations on inhibitor testing – lessons from External Quality Assessment (EQA) (EJF, BV) Nevertheless, inter-laboratory coefficients of variation (CV) approaching (and sometimes exceeding) 50% are evidenced by various EQA groups, inclusive of ECAT (External quality Control of Assays and Tests) Foundation, NEQAS (UK National External Quality Assurance Scheme) and RCPA (Royal College of Pathologists of Australasia) (Fig. 1) [7–10]. This variability has not improved over the years. Reasons for such high variability largely derive from assay © 2014 John Wiley & Sons Ltd
INHIBITOR TESTING (a)
95
(b)
Fig. 1. Summary of recent EQA data highlighting inter-laboratory variation for factor inhibitors. (A) Data for FVIII inhibitor testing using composite data for RCPAQAP showing coefficient of variation (CV, %; y-axis) plotted against median participant reported inhibitor level (in Bethesda units [BU]/mL; x-axis) for all samples distributed over the past 6 years, and shown separately for Bethesda or Nijmegen methods. The average and 95% confidence intervals of the mean of all data are shown on the right of the figure. (B) Recent data for ECAT for FIX inhibitor testing showing the participant reported inhibitor level (y-axis; BU/mL) for two samples distributed in 2013 (identified on x-axis). Figure redrawn from reference 10.
variability, which can occur at any stage within the assay, including choice and source of laboratory reagents, buffering (yes, no, how), NPP, etc., such that no two laboratories participating in the RCPA EQA programme could be found to be using exactly the same procedure in one analysis [8]. Although NA tend to provide lower variability than BA [7–10], how laboratories perform NA varies almost as much as how laboratories perform BA; thus, variability still remains high. The percentage of falsely positive and falsely negative results in FVIII-inhibitor-negative samples is also unacceptably high (up to 32% for false positives; up to 5% for false negatives) [7–10]. To improve reproducibility, ECAT performed a quality improvement cycle including these steps: (i) an external survey among 51 laboratories that participated on a regular basis in the ECAT FVIII-inhibitor programme; (ii) selection from these of 15 representative laboratories for a centralized workshop; (iii) during the workshop, a zero point measurement using participants’ own methods and reagents and (iv) separate measurements with a fully standardized and universal method (NA) and reagents; (v) an external survey among 13 workshop participants 3 months after the initial workshop and (vi) an external survey among 22 of the original 51 laboratories with standardized methods (BNPP with FVIII activity ranging from 0.95 to 1.05 IU mL 1, FVIII-deficient plasma as reference sample, standardized sample dilution with FVIII-deficient plasma). In each step of the cycle, an identical set of seven samples was used (one negative and six positive). The means and CVs of results using the six inhibitorpositive samples at the various steps (Table 1) clearly show that very low inter-laboratory variations can be achieved using a centralized setting with uniform methods and reagents (step 4), and acceptable inter© 2014 John Wiley & Sons Ltd
laboratory variations are possible in EQA surveys by significant standardization of the methods (step 6). The inhibitor activity of the negative sample was also below the cut-off value in all participants bar one in steps 4 and 6. Additional pitfalls as well as strategies for improvements in this area of testing are extensively covered in a recent review [10]. In brief, pitfalls occur at any stage of the diagnostic process, including pre-analytical issues (doctors test request, sample collection), analytical (methodology as detailed above), and post-analytical (laboratory interpretation and reporting, doctors’ interpretation and action). Main strategies comprise: (a) pre-analytical: (i) check test orders for accuracy and relevance (to ensure performance of the correct tests); (ii) check and be aware of blood sampling issues (correct anticoagulant, proper fill, etc.; this will help avoid inaccuracies, and false positive/negative test findings); (b) analytical: (i) be aware of assay limitations such as low level inhibitor sensitivities, (ii) use the best methodologies and implement proven assay improvements (helps ensure accuracy), (iii) perform appropriate internal quality control and participate in EQA (helps ensure accuracy), (iv) repeat tests when necessary (helps ensure accuracy and avoid false negatives/positives); (c) post-analytical: (i) request repeat testing using fresh samples for confirmation when test results do not match expectations; (ii) provide consultative services with clinicians to assist in test interpretation.
Further enhancing sensitivity and specificity of inhibitor testing (BV) It may be concluded that implementation of more standardized methods will lead to better specificity, as Haemophilia (2014), 20 (Suppl. 4), 94--98
96
E. J. FAVALORO et al.
Table 1. Recent findings from an ECAT study.
Sample number (nominal inhibitor activity) 1 (1.6 BU mL 1) 2 (0.8 BU mL 1) 3 (1.4 BU mL 1) 4 (0.7 BU mL 1) 5 (1.9 BU mL 1) 6 (15.4 BU mL 1) Mean CV
Steps 1 + 2
Steps 3 + 4
Pre-workshop survey (2009)
Workshop results (2009)
Step 5
Step 6
Post-workshop survey (2010)
Standardized final survey (2012)
51 Laboratories
15 laboratories selected for the workshop
Zero measurement
Standardized measurement
13 Laboratories
22 Laboratories
2.32 (36%) 0.79 (49%) 0.9 (41%) 0.44 (70%) 1.74 (36%) 11.0 (36%) 45%
2.69 (43%) 1.02 (31%) 1.16 (39%) 0.59 (69%) 1.74 (37%) 11.5 (44%) 44%
2.97 (39%) 1.33 (69%) 1.17 (30%) 0.61 (45%) 2.34 (41%) 14.9 (41%) 44%
1.93 (8%) 0.94 (5%) 1.16 (6%) 0.50 (13%) 2.22 (12%) 14.6 (6%) 8%
2.9 (41%) 1.1 (88%) 1.1 (31%) 0.6 (61%) 1.9 (31%) 12.0 (36%) 48%
2.7 (31%) 0.7 (17%) 1.0 (23%) 0.5 (30%) 1.8 (22%) 12.4 (27%) 25%
Mean inhibitor activity in BU mL 1; inter-laboratory coefficient of variation (CV) given in brackets.
evidenced by the above ECAT study. Besides problems with reproducibility and specificity, both BA and NA lack sensitivity for low inhibitor activities. The internationally agreed detection limit for both tests is 0.6 Bethesda units (BU), although it may be lower for NA because of improved specificity. Nevertheless, both assays may miss inhibitors with low activity. From personal experience, it was hypothesized that these lowtitre (‘undetectable’) inhibitors might be clinically significant and present in patients in the late Immune Tolerance Induction (ITI) phase, rendering replacement therapy less effective and leading to bleeding complications and increased need of FVIII concentrates in these patients. Therefore, a low-titre FVIII inhibitor assay (LTA) was recently developed and described with a lower limit of detection of 0.03 BU [11]. The principle of the LTA is identical to NA except for the use of concentrated plasma instead of native plasma, an alternative ratio of concentrated plasma/BNPP in the test mixture of 3:1, and the use of chromogenic substrates for assay of residual FVIII. Assay results are expressed in BU by correcting the analysis data for the concentration factor of the plasma and the alternative ratio. Using LTA, low-titre inhibitors were demonstrated as still present in the early post-ITI phase in haemophiliacs treated for FVIII inhibitors despite negative findings with NA or BA. These low-titre inhibitors decrease the half-life and the recovery of infused FVIII products [11]. The clinical significance of LTA was further evaluated in a satellite study of the International ITI (I-ITI) study [12], in which inhibitor-positive patients were treated with low (50 IU kg 1 thrice weekly) or high (200 IU kg 1 per day) dose regimens of recombinant FVIII concentrates until the NA or BA became negative (
