Performance characteristics of 5 numerical indexes in mixing test interpretation under coexistence of lupus anticoagulant and coagulation factor deficiency

Abstract

Background

The mixing test is useful to investigate the cause of unexpectedly prolonged activated partial thromboplastin time (APTT). Several indexes are available for distinguishing correction from non-correction (ie, factor deficiency from inhibitors), but their performance characteristics may differ because of their different formulas. Furthermore, it is unclear how each index performs under the coexistence of factor deficiency and inhibitors.

Objectives

The objective of this study was to examine the differences in indexes, depending on factor VIII activity (FVIII:C) levels and lupus anticoagulant (LA) titers in test samples.

Methods

APTT was measured in spiked samples with various FVIII:C levels and LA titers, normal pooled plasma (NPP), and their 4:1, 1:1, and 1:4 mixtures. The following 5 indexes were calculated: index of circulating anticoagulant, mixing test normalized ratio, 4:1 and 1:1 percent corrections, and an APTT difference between the 1:1 mixture and NPP. The samples with LA, showing correction, were measured for FVIII:C in a one-stage assay to check parallelism.

Results

All indexes showed correction under FVIII deficiency and non-correction under higher LA titers. However, under lower LA titers, some indexes showed non-correction but others showed correction because of dilution effects and variations in formulas and/or sample mix ratios. The differences among the indexes were more pronounced under coexistent FVIII deficiency and LA, even though LA titers were equal in the tested samples; samples with lower FVIII:C showed correction, whereas those with normal FVIII:C showed non-correction. The samples tested for FVIII:C showed non-parallelism.

Conclusion

Each index had different performance characteristics to LA samples, which were pronounced under low FVIII:C levels in test samples.

Essentials

• •Mixing tests may distinguish coagulation inhibitors from coagulation factor deficiency.
• •Mixing test performance with coexistent lupus anticoagulant (LA) and factor deficiency is unclear.
• •Performance regarding LA detection varied among the evaluated correction indexes.
• •Differences in the indexes were more pronounced with coexistent LA and greater factor deficiency.

1. Introduction

The mixing test is a simple but practically powerful tool to investigate the cause of unexpectedly prolonged clotting time (CT) (eg, activated partial thromboplastin time [APTT]) [[1], [2], [3]]. Patient plasma is mixed with normal pooled plasma (NPP), and CT on the mixture is measured to assess whether the original prolonged CT becomes normal (corrected) or remains prolonged (non-corrected). Correction generally indicates coagulation factor deficiency, whereas non-correction suggests the presence of an inhibitor (eg, lupus anticoagulant [LA] or a specific coagulation factor inhibitor) [3]. Distinction of correction/non-correction is the principal purpose of the mixing test, guiding follow-up tests (eg, coagulation factor assays or LA test) [1]. Therefore, the mixing test is incorporated as part of the strategy to identify the prolonged CT cause; usually performed before follow-up tests because they are costly and time-consuming [1,3].

Because the correction/non-correction interpretation is subjective, numerical indexes are reportedly useful in observed correction [[1], [2], [3], [4], [5]]. There are several available indexes as follows: index of circulating anticoagulant (ICA), mixing test normalized ratio, percent correction, and CT difference between 1:1 mixture and NPP. Since the indexes differ in their formulas and the mix ratio of patient plasma to NPP [[1], [2], [3], [4], [5]], the performance characteristics, namely the sensitivity and specificity to detect factor deficiency and inhibitors, may vary among the indexes. For instance, ICA has a high specificity to factor deficiency, whereas the mixing test ratio has a high sensitivity to inhibitors [6,7], although both indexes are recommended for LA diagnosis [8,9]. Previous studies have improved the sensitivity and specificity by establishing suitable cutoff values [1,6].

The simple purpose of the mixing test raises the question of how each index performs under the coexistence of factor deficiency and inhibitors. Regarding this, there are several clinical cases, such as hemophilia with allo- or auto-antibody, hemophilia with LA, and multiple factor deficiency (eg, liver dysfunction or vitamin K antagonist [VKA] therapy) with LA [1,10,11], although LA testing in patients receiving VKA is not recommended [12]. The first case can be overcome by incubating the mixture and then measuring its APTT [[1], [2], [3]], but in other cases, factor deficiency or inhibitor (ie, LA) can be detected, but not both, because of the mixing test principle of distinguishing correction/non-correction. Theoretically, the mixing tests would show non-correction [2] because despite adding NPP to overcome factor deficiency, the presence of LA persists. However, this is not always the case [10,11]. A previous study examining patients with both hemophilia and LA, in which mixing tests were performed without incubation, showed split results (non-correction in some patients but not in others) [11]. One primary reason could be the variety of numerical indexes. Although performance characteristics in each index have been well examined in samples with a single cause of APTT prolongation [1,[5], [6], [7]], those with 2 causes (eg, factor VIII [FVIII] deficiency and LA) remain largely unknown.

Our study aimed to examine performance characteristics of 5 numerical indexes in APTT mixing tests in spiked samples that are FVIII-deficient and LA-positive because they have been known to show conflicting results in APTT mixing tests. FVIII-deficient samples show correction and LA-positive samples show non-correction. Hence, the use of spiked samples is advantageous in preparing varying levels of FVIII activity (FVIII:C) and LA titers. We expect to reveal each index characteristic, depending on the FVIII:C levels and LA titers in the test samples.

2. Materials and Methods

2.1. Plasma samples

Commercially available NPP (CRYOcheck Pooled Normal Plasma) and FVIII-deficient plasma (CRYOcheck Factor VIII Deficient Plasma) both from Precision BioLogic were used; their measured FVIII:C was 96.9 IU/dL and less than 1 IU/dL, respectively. To prepare serially FVIII:C-decreasing samples, NPP and FVIII-deficient plasma were mixed in ratios of 1:1, 1:3, 1:7, and 1:15; their measured FVIII:C was 48.6, 24.0, 12.4, and 6.4 IU/dL, respectively. To generate sequential LA titers, we used a commercially available mouse monoclonal immunoglobulin G (IgG) antibody against phosphatidylserine/prothrombin complex (anti-PS/PT) (Medical and Biological Laboratories [MBL]) [13]. Anti-PS/PT is a major antibody responsible for LA [14], causing APTT prolongation [13]. The commercial anti-PS/PT (1 mg/mL) and its diluted samples (0.75 and 0.50 mg/mL) were mixed with NPP and prepared FVIII:C-decreased samples (1:19). BALB/c mouse IgG (Molecular Innovation) was used as a control and added in a similar manner because the commercial anti-PS/PT was established using BALB/c mice [13]. In total, we prepared 20 samples by combining 4 levels of anti-PS/PT concentrations (ie, LA titers) with 5 levels of FVIII:C, in which the final levels for anti-PS/PT concentrations were 0, 25, 37.5, and 50 μg/mL, and for FVIII:C were 92.1, 46.2, 22.8, 11.8, and 6.1 IU/dL. The LA titers were also represented as the screen/confirm ratio in dilute Russell’s viper venom time (dRVVT) test [8,9] (cutoff ≥1.3, LA test Gradipore, MBL), and the values measured were 1.10, 1.47, 1.58, and 1.75.

The study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of Niigata University, Japan (no. 2018-0316).

2.2. APTT mixing tests and interpretation

APTT mixing tests, without incubation, were performed in duplicate in the prepared samples, using an automated coagulation analyzer (CP3000, SEKISUI MEDICAL) and an LA-sensitive reagent, Thrombocheck APTT-SLA (Sysmex) [6]. The mix ratios of the test plasma to NPP were 1:1, 4:1, and 1:4, as described in previous studies [1,4,5]. This study introduced 5 indexes to interpret the mixing test results: ICA, mixing test normalized ratio, 1:1 and 4:1 percent corrections, and Mix–NPP (Table 1), in which the cutoff values were followed as described in previous studies [1,5,6,15].

Table 1.

Profile of indexes used for interpretation of APTT mixing tests.

Index	Formula	Cutoff value
ICA	[(c-d)/a] × 100	12.4
Mixing test normalized ratio	c/d	1.12
1:1 percent correction	[(a-c)/(a-d)] × 100	72
4:1 percent correction	[(a-b)/(a-d)] × 100	50
Mix–NPP	c-d	8

Each index was calculated, in which a, b, c, and d represent the APTT values (seconds) of neat test plasma, mixture of 4 parts test plasma to 1-part NPP, 1:1 mixture, and NPP, respectively. d represents average in repeated measurements of NPP (n = 40). The cutoff values used were determined as described in previous studies.

Abbreviations: APTT, activated partial thromboplastin time; ICA, index of circulating anticoagulant; Mix, 1:1 mixture; NPP, normal pooled plasma.

2.3. One-stage FVIII assays

One-stage FVIII assays were performed on samples containing anti-PS/PT and showing correction in mixing tests, using multiple sample dilutions doubling from 1/5 to 1/80, to check parallelism [16,17]. The samples containing an inhibitor (eg, LA) showed falsely decreasing and higher coagulation factor activity values in lower and higher dilution samples, respectively, which is well known as non-parallelism, whereas the samples containing no inhibitors showed similar values irrespective of dilution ratios [16]. LA-positive is a typical case showing non-parallelism [16]. In one-stage FVIII assays, APTT reagent, FVIII-deficient plasma, and calibrator to generate a standard curve were Thrombocheck APTT-SLA, Thrombocheck Factor VIII (both from Sysmex), and Standard Human Plasma (Siemens), respectively.

3. Results and Discussion

Differences in formulas and mix ratios of test samples to NPP in each index were observed as anticipated, especially in LA samples. In samples with lower FVIII:C levels without anti-PS/PT, all indexes showed correction (Figure 1), indicating that each index has good performance characteristics to factor deficiency. By contrast, in LA samples, the results differed among the indexes (Figure 1), although the mixing tests should show non-correction theoretically [2]. The mixing test normalized ratio showed non-correction in all LA samples, indicating that this index is most sensitive to inhibitors, which is consistent with the previous studies and LA diagnosis guidelines in which mixing test-specific cutoff was used [[6], [7], [8], [9]]. However, the other indexes showed split results depending on LA titers. The ICA and Mix–NPP showed correction in lower LA samples; both indexes calculate the APTT difference between mixture and NPP in their formulas. A possible reason would be the dilution effect [18], namely, mixing with NPP diluted LA (ie, anti-PS/PS concentration), causing small APTT prolongation. In fact, the APTT in mixtures was within the normal range [19] (Figure 2B). Nonetheless, the 1:1 and 4:1 percent correction indexes showed non-correction in lower LA samples under normal FVIII:C levels; both indexes calculate APTT difference between neat test sample and mixture in their formulas. The differences in each index formula may contribute to performance characteristics, although cutoff values were crucial in the judgment. Interestingly, the 4:1 percent correction index identified more samples as non-correction than 1:1 percent correction (Figure 1). The difference between the 2 indexes lies in the mix ratio of the test sample to NPP, in which 4:1 percent correction uses more test sample. The difference is pivotal because the mixture containing more test samples also showed more prolonged APTT (Figure 2). Hence, hemostasis laboratories need to understand the performance characteristics of each index before using it because no indexes permit 100% accurate discrimination [3].

Figure 1.

Comparison of 5 indexes in APTT mixing test interpretation in samples with various levels of LA titers and FVIII:C. The 20 samples, combining 4 levels of anti-PS/PT concentration (ie, LA titers) with 5 levels of FVIII:C, were tested. The mixing test results in each sample are represented as calculated numerical indexes of ICA (A), mixing test normalized ratio (B), 1:1 percent correction (C), 4:1 percent correction (D), and Mix–NPP (E). The Mix–NPP is the difference in APTT (seconds) between 1:1 mixture and NPP. The cutoff values (dashed lines) in each index were determined as described in previous studies. The judgment of non-correction (ie, LA detection) depended on FVIII:C in neat test samples, despite equal LA titers in samples tested; lower FVIII:C (lighter blue bar) levels caused the decision of “factor deficiency.” anti-PS/PT, antibody against phosphatidylserine/prothrombin complex; APTT, activated partial thromboplastin time; Factor def, coagulation factor deficiency; FVIII:C, coagulation factor VIII activity; ICA, index of circulating anticoagulant; LA, lupus anticoagulant; Mix, 1:1 mixture; NPP, normal pooled plasma.

Figure 2.

APTT for each ratio of test sample to NPP in the mixtures in APTT mixing tests. In this mixing test study, APTT was measured in neat test sample, NPP, and the mixture of the 2. The mix ratio was 1:4, 1:1, and 4:1. The measured APTT was plotted graphically on the Y-axis, and the mix ratio of test sample in the mixture along the X-axis, according to the anti-PS/PT concentration of 0 (A), 25 (B), 37.5 (C), and 50 μg/mL (D). The dashed lines represented the upper limit of NRI in the APTT reagent used, established in our laboratory. anti-PS/PT, antibody against phosphatidylserine/prothrombin complex; APTT, activated partial thromboplastin time; FVIII:C, coagulation factor VIII activity; NPP, normal pooled plasma; NRI, normal reference interval.

Notably, under the coexistence of FVIII deficiency and LA, FVIII:C in test samples largely affects LA detection in most indexes. Even though the LA titers were equal in samples tested, the samples with lower FVIII:C were identified as correction, whereas those with normal FVIII:C were non-correction (Figure 1), although only the mixing test normalized ratio found non-correction in all LA samples irrespective of the FVIII:C. The difference between the mixing test normalized ratio and other indexes such as ICA and percent correction is whether the CT in neat test sample is used for their calculation. This would be pivotal because the CT in neat sample was prolonged by both factor deficiency and inhibitors, largely influencing the calculated index values. Our results also suggest that coagulation factor deficiency strongly influences the results of mixing tests [10]. The impact of FVIII:C in LA detection was more pronounced in 1:1 and 4:1 percent correction indexes than that of ICA. These indexes showed a dynamic decrease in their values, depending on the FVIII:C levels in test samples (Figure 1), indicating that more careful attention is needed in the interpretation using the indexes. Owing to the coexistence of LA and coagulation factor deficiency of FVIII and other coagulation factors, such as factor X [20], it is important to recognize that LA detection in APTT mixing tests can be influenced by coagulation factor activity levels in test samples, and the impact may differ in indexes.

Because APTT mixing tests judge correction/non-correction, it is important to establish a subsequent testing strategy to ensure that undetected cause is also tested for, considering the possibility of coexistence of factor deficiency and inhibitors. The correction does not rule out the presence of inhibitors, considering the dilution effect [18], and the non-correction does not rule out factor deficiency because of the addition of NPP to test plasma in the testing procedure [2]. When the mixing tests show correction, coagulation factor measurements as follow-up tests are required [1], wherein the recommended subsequent strategy is to use a one-stage assay using multiple sample dilutions to check parallelism (data for some samples were shown in Figure 3). The assay can check the presence of inhibitors while simultaneously measuring factor activity [16]. Although multiple dilutions are strongly recommended in the best practice guidelines [21], this might not always be the default position [22], including in Japan. Our results also recommend using multiple dilutions, in which additional lower dilution ratios such as 1/5 might be helpful to reliably and easily read the non-parallelism (Figure 3). On the other hand, for samples showing non-correction, investigation of inhibitors (eg, dRVVT test) and factor activities using chromogenic assays is recommended because these assays are not affected by inhibitors [23]. Hence, the mixing tests would help guide the next steps.

Figure 3.

One-stage FVIII assays in samples with various levels of LA titers and FVIII:C. In one-stage FVIII assays, multiple sample dilution, doubling from 1/5 to 1/80, was performed to check parallelism (or non-parallelism). LA titers were represented as anti-PS/PT concentration. Non-parallelism, meaning falsely lower FVIII:C levels in lower dilution samples and higher FVIII:C levels in higher dilution samples, was identified in samples with anti-PS/PT (green lines), but not in those without anti-PS/PT (gray line). Non-parallelism was identified in LA samples with not only normal FVIII:C levels but also lower FVIII:C levels. anti-PS/PT, antibody against phosphatidylserine/prothrombin complex; FVIII:C, coagulation factor VIII activity; LA, lupus anticoagulant.

This study has some limitations. First, only spiked plasma samples were tested. Second, we used only one combination of reagents and analyzer; the cutoff values in each index may depend on the combination used. These results should be further validated using clinical samples.

In conclusion, this study revealed characteristics of each index depending on the levels of FVIII:C and LA titers in test samples. Although every index identified FVIII-deficient samples as correction, split results of correction/non-correction among the indexes were found in LA samples, which was more pronounced under the coexistence of lower FVIII:C levels. These findings would be clinically crucial for patients with both LA and lower factor levels. LA is identified in patients with antiphospholipid syndrome [14,24] and also other diseases such as coronavirus disease 2019 [25], in which anti-PS/PT is detected [26]. Lower coagulation factor levels are identified in congenital hemophiliacs (including carriers), and other acquired disorders [20]. The APTT mixing test is useful to delineate the 2 factors, but it potentially becomes a disadvantage in samples that have both factors. Therefore, APTT mixing tests will be useful in identifying subsequent testing procedure rather than determining just one cause.