ORIGINAL ARTICLE A new bead-based human platelet antigen antibodies detection assay versus the monoclonal antibody immobilization of platelet antigens assay Leendert Porcelijn,1 Elly Huiskes,1 Ilona Comijs-van Osselen,1 Aniska Chhatta,1 Vipul Rathore,2 Matthew Meyers,3 and Masja de Haas1
BACKGROUND: The performance of a newly developed Luminex bead-based platelet (PLT) antibody detection method (PAKLx) was compared with the monoclonal antibody immobilization of PLT antigens (MAIPA) assay and the LifeScreen Deluxe Luminex bead-based HLA Class I antibody detection method (LMX). STUDY DESIGN AND METHODS: Six sera containing anti-human PLT antigen (HPA)-1a (n = 2), HPA-1b, HPA-2b, HPA-3a, or HPA-5b were tested in titration. A total of 194 sera, including HPA-1a, -1b, -2a, -2b, -3a, -5a, and -5b antibodies with or without HLA antibodies (n = 63); glycoprotein (GP) IV antibodies (n = 1); PLT autoantibodies (n = 3); HLA antibodies (n = 45); and samples with no PLT-reactive antibodies (n = 82), were tested in both assays. RESULTS: Comparable levels of sensitivity were obtained for the MAIPA and PAKLx. The PAKLx showed four (6%) false-negative results in 67 sera with HPA or GP-reactive antibodies: anti-HPA-3a (n = 1) or antiHPA-5b (n = 3). The PAKLx showed in 10 of the total 194 samples (5%) the presence of antibodies not detected by the MAIPA. This concerned broadly GP-reactive antibodies (n = 7), anti-GPIIb/IIIa combined with anti-HPA-3a (n = 1), anti-HPA-1a (borderline, n = 1), and anti-GPIV (n = 1). Testing 175 sera for antiHLA Class I antibodies in the PAKLx and LMX showed four discrepant results: PAKLx negative and LMX positive, n = 3 and n = 1, respectively. CONCLUSION: For the vast majority of the specimens tested (93%) the results of the PAKLx were in concordance with the MAIPA. The PAKLx is a fast, easy to perform, and sensitive PLT antibody screening method.
A
lloantibodies directed against human platelet antigens (HPA) are of clinical significance in fetal and neonatal alloimmune thrombocytopenia (FNAIT),1 refractoriness to platelet (PLT) transfusions,2 and posttransfusion purpura.3 Currently, several techniques are used for the detection of PLT-specific alloantibodies, such as the PLT immunofluorescence test (PIFT),4 the monoclonal antibody immobilization of PLT antigens (MAIPA) assay,5 and various types of solid-phase enzyme-linked immunosorbent assay using PLT-specific glycoproteins (GPs).6,7 Until now the MAIPA is the most sensitive and specific PLT alloantibody detection technique.8 The MAIPA is a laborious assay that takes 6 to 8 hours and due to the necessity to use HPA-typed PLTs; it can only be performed in specialized reference laboratories.
ABBREVIATIONS: FNAIT = fetal and neonatal alloimmune thrombocytopenia; GP = glycoprotein; LIFT = lymphocyte immunofluorescence technique; LMX = LifeScreen Deluxe Luminex bead-based HLA Class I antibody detection method; MAIPA = monoclonal antibody immobilization of platelet antigens; PIFT = platelet immunofluorescence test. From the 1Sanquin Diagnostic Services, Amsterdam, The Netherlands; and 2Immucor-GTI Diagnostics and 3HOLOGIC, Waukesha, Wisconsin. Address reprint requests to: Leendert Porcelijn, MD, Department of Immunohematology Diagnostic Services, Platelet/Leukocyte Serology Laboratory, Sanquin Diagnostic Services, Plesmanlaan 125, 1066CX Amsterdam, The Netherlands; e-mail: [email protected]. The PAKLx tests were provided by Gen-Probe Diagnostics, Inc. and PAKLx materials were provided by Lifecodes products, HOLOGIC/Gen-Probe. Received for publication July 11, 2013; revision received October 4, 2013, and accepted October 22, 2013. doi: 10.1111/trf.12509 © 2013 AABB TRANSFUSION **;**:**-**. Volume **, ** **
TRANSFUSION
1
PORCELIJN ET AL.
Bead-based HPA antibody detection assays for only the detection of anti-HPA-1 have been described by three different groups. Bakchoul and colleagues in 20079 used purified GPIIb/IIIa from outdated PLTs in a gel antigenspecific assay, Chong and colleagues in 201010 used recombinant β3-integrins displaying the HPA-1a or HPA-1b epitope, and Skaik and colleagues in 201311 used recombinant soluble β3-integrins displaying HPA-1a or HPA-1b epitopes in a single-antigen magnetic bead assay. Fujiwara and coworkers described in 200912 the immunecomplex capture fluorescence analysis, a kind of beadbased MAIPA in which they used microarray beads coupled with monoclonal antibodies (MoAbs) specific for CD36, CD41, CD42b, CD49b, CD61, and HLA Class I antigens to capture GP-antibody complexes. Recently, the PAKLx bead-based GP-specific HPA antibody detection method (Lifecodes products, HOLOGIC/Gen-Probe, Waukesha, WI) became available. This assay uses beads to which PLT lysate–derived GPs carrying clinical relevant HPA or HLA Class I specificities are coupled to capture and identify HPA and HLA Class I antibodies. This assay is easy to perform in 3 hours and uses only 10 μL of serum. We compared MAIPA and PAKLx test results for 194 serum samples from women who gave birth to a child suspected to suffer from FNAIT (n = 100), patients with refractoriness for PLT transfusions (n = 74), and healthy male donors (n = 20).
MATERIALS AND METHODS Patient samples Six sera with HPA antibodies of different specificities (see Table 1) were tested in titration in the MAIPA and
Antibody detection assays
TABLE 1. PAKLx test beads Sample 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
2
GP group Control bead 1 Control bead 2 Control bead 3 Positive control bead GPIV HLA Class I GPIIb-IIIa GPIIb-IIIa GPIIb-IIIa GPIIb-IIIa GPIIb-IIIa GPIIb-IIIa GPIb-IX GPIb-IX GPIb-IX GPIb-IX GPIb-IX GPIa-IIa GPIa-IIa GPIa-IIa GPIa-IIa GPIa-IIa
TRANSFUSION Volume **, ** **
PAKLx. A total of 100 sera from women who gave birth to a child suspected to suffer from FNAIT, 74 sera from patients who showed PLT refractoriness to PLT transfusions (refractoriness cases), and 20 sera from healthy male (never transfused) donors were tested in the MAIPA and PAKLx. Serum samples from 174 cases were referred to the laboratory of Platelet and Leukocyte Serology of Sanquin Diagnostic Services for FNAIT or refractoriness to PLT transfusion investigation. One or more anti-HPA and/or GP-reactive antibodies (with or without HLA antibodies) were present in 67 of the 174 samples (as detected in the MAIPA); anti-HPA-1a (n = 29; optical density [OD] in MAIPA varying from 0.5 to >3.0 [cut-off level OD, 0.15]), anti-HPA-1b (n = 5; OD in MAIPA varying from 0.15 to >3.0), anti-HPA-2a (n = 1; OD, 2.5), anti-HPA-2b (n = 4; OD varying from 0.25 to 1.2), antiHPA-3a (n = 4; OD varying from 0.4 to >3.0 for three of these antibodies and one labile anti-HPA-3a only detectable in the PIFT as described by von dem Borne et al.4 and in the MAIPA if using some other, not routinely used, MoAbs, see below), anti-HPA-5a (n = 6; OD varying from 0.4 to 2.3), anti-HPA-5b (n = 16; OD varying from 0.2 to 2.5), autoantibodies (n = 3; two samples scored as weak positive in MAIPA [OD, 0.2-0.4] with CD49b/CD29 [GPIa/ IIa], CD42 [GPIb/IX], and CD41/CD61 [GPIIb/IIIa] and one sample scored as weak positive in MAIPA [OD, 0.2] with GPIa/IIa and GPIb/IX), and anti-GPIV (CD36, n = 1; not detectable in the MAIPA, but reactive in the PIFT with a panel of typed donor PLT suspensions). Of the 174 collected samples, according to the regulation of Sanquin Diagnostic Services, anonymized leftover materials were used. Written informed consent was obtained for the 20 donor samples.
Antigens
GPIV HLA Class I HPA-1a-3a-4a HPA-1a-3b-4a HPA-1b-3a-4a HPA-1b-3b-4a HPA-1ab-3ab-4a HPA-1a-3ab-4b HPA-2a HPA-2a HPA-2ab HPA-2b HPA-2b HPA-5a HPA-5a HPA-5ab HPA-5b HPA-5b
The PAKLx (Lifecodes products, HOLOGIC/Gen-Probe) is a bead-based technique for detection and specification of antibodies against antigens located on GPIIb/IIIa—HPA1a, -1b, -3a, -3b, -4a, and -4b; on GPIb/IX—2a and 2b; on GPIa/IIa—5a and 5b; GPIV (beads coated with affinity purified GPs from HPA typed PLT donors); and HLA Class I (bead coated with affinity purified HLA Class I obtained from a PLT pool that was composed of 100 Caucasian, 100 African American, and 100 Hispanic blood donors; Table 1). A kit is supplied containing 800 μL of beads. The beads include a positive control, three negative control beads, and 18 HPA or HLA Class I–specific beads, 100 μL of phycoerythrin (PE)-conjugated goat anti-human immunoglobulin (Ig)G, 1 mL of diluent, 30 mL of phosphate-based wash buffer, 80 μL of positive control serum (anti-HPA-1a), and 80 μL of negative control serum (inert serum). Forty microliters of pooled beads is incubated with 10 μL of patient serum for 60 minutes and washed four times with wash buffer, after which 50 μL of
BEAD-BASED HPA ANTIBODY DETECTION
PE-labeled anti-human IgG is added. After 30 minutes of incubation, antibody binding is measured using a Luminex instrument. Specific software is provided to determine if a result is positive. The mean fluorescence intensity (MFI) of the individual bead is divided by the MFI for each negative control, after which a background adjustment factor is subtracted for the appropriate bead/ control bead combination (bead MFI/negative control bead 1 MFI minus background adjustment factor = adjusted ratio 1, idem for 2 and 3). If two or three of the adjusted ratios are positive, the result is positive. The MAIPA used is a modification of the MAIPA described by Kiefel and colleagues.5 In short, a panel of typed donor PLTs (15 × 106 if testing for GPIIb/IIIa and 40 × 106 for the other GP) were incubated with 120 μL of patient serum (maternal sera for FNAIT cases) and a MoAb (all from our institution, Division Reagents of Sanquin Blood Supply) recognizing GPIIb/IIIa (CLBthromb/1, CD41, anti-GPIIb), GPIb/IX (MB45, CD42a, anti-GPIX), GPIa/IIa (10G11, CD49b, anti-GPIa), GPIV (P58, CD36), or HLA Class I (W6.32). For the detection of one labile anti-HPA-3a (see below) we also used the GPIIb/IIIa-specific MoAb Y2/51 and PL164-A101 (CD61 and CD41, MoAb workshop Vienna, 1989) and CLBthromb/7 (CD41, Division Reagents of Sanquin Blood Supply). After binding of the MoAb, the PLTs were solubilized with an Nonidet P-40–containing buffer. The lysates were added to a microtiter plate coated with goat immunoglobulin anti-mouse IgG to immobilize the MoAb-GP-human antibody complexes. Antibodies bound to the analyzed GP or GP-complex are detected with horseradish peroxidase labeled goat immunoglobulin anti-human IgG. The samples were tested for antibodies against HPA-1a, -1b, and -3a (GPIIb/IIIa); HPA-2a and 2b (GPIb/IX); HPA-5a and -5b (GPIa/IIa); and HLA Class I and GPIV. Discrepant results were retested twice in the PAKLx to finally use the results detected in two of three tests, and once in the MAIPA to confirm the initial MAIPA results to exclude sample mix-up or loss or gain of signals. Because the MAIPA is not routinely used for the detection of Class I HLA-reactive antibodies, the LifeScreen Deluxe (LMX) assay, a bead-based immunoassay, was used for the detection of HLA-reactive antibodies.13 All 194 samples used in the comparison were first qualified for the study by performance in LMX. Any sample that produced indeterminate, borderline, and/or ambiguous results in LMX were excluded from the study for consideration, leaving 175 LMX well-defined samples. These 175 samples were tested both in the PAKLx with the HLA Class I–carrying bead (coated with affinitypurified HLA Class I GPs obtained from PLTs of 100 Caucasian, 100 African American, and 100 Hispanic blood donors) and with the comparable HLA Class I–coated bead in the LMX.
Statistical analysis Correlations between MAIPA ODs and PAKLx MFIs were calculated with computer software (Microsoft Excel 2002, Microsoft Corp., Redmond, WA).
RESULTS Anti-HPA-1a (n = 2), anti-HPA-1b, anti-HPA-3a, and antiHPA-5a were all correctly detectable in the first test of six selected sera with the PAKLx. Tested in dilution, sera containing anti-HPA-1a (n = 2), anti-HPA-1b (n = 1), anti-HPA-3a (n = 1), and anti-HPA-5a (n = 1) all showed comparable results in the MAIPA and PAKLx (Table 2). Anti-HPA-2b was detectable up to a dilution of 32 in the MAIPA compared to a dilution of 8 in the PAKLx. Of the 100 maternal sera from FNAIT cases, 39 showed negative results both in the MAIPA and in the PAKLx (Table 3). In one of these 39 samples an antiHPA-3a was detected in the PIFT. This HPA-3a antibody was not detected by PAKLx or by our routine MAIPA using the MoAb CLBThromb/1. However, it did react in the MAIPA using the MoAb Y2, 6C9, and P21 (OD varying from 0.16 to 0.90). Forty-four sera reacted positive in both techniques. In these sera the following antibody specificities were found: anti-HPA-1a with or without anti-HLA (n = 17), antiHPA-1b with anti-HLA (n = 1), anti-HPA-5a with or without anti-HLA (n = 4), anti-HPA-5b with or without anti-HLA (n = 10), anti-HPA-1a with -5b with anti-HLA (n = 1), anti-HPA-3a with -5b with anti-HLA (n = 1), and anti-HLA (n = 10). The 17 samples showing discrepant results between MAIPA and PAKLx are shown in Table 4. Ten samples showed results discrepant for the detection of GP or HPA-reactive antibodies between MAIPA and
TABLE 2. Six sera containing antibodies with a HPA specificity tested in titration MAIPA results HPA-1a HPA-1a HPA-1b HPA-2b HPA-3a HPA-5b
MAIPA titer
PAKLx results
PAKLx titer
>128* 256 32 32 1024 1024
Anti-HPA-1a Anti-HPA-1a Anti-HPA-1b Anti-HPA-2b Anti-HPA-3a Anti-HPA-5b
>128* 256 32 8 1024 1024
* The serum was not further diluted.
TABLE 3. Antibody detection in samples from possible FNAIT cases (n = 100) PAKLx MAIPA Negative Positive
Negative
Positive
39 1
3 44 + 13*
* Thirteen discrepant positive results in MAIPA and PAKLx.
Volume **, ** **
TRANSFUSION
3
PORCELIJN ET AL.
TABLE 4. Discrepant results obtained with sera from cases analyzed because of possible FNAIT* Sample 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
MAIPA Negative Anti-HLA Anti-HPA-1a Anti-HPA-1a Anti-HPA-1a Anti-HPA-1a Anti-HPA-1a + 3a (OD 0.55) Anti-HPA-5b (OD 0.55) Anti-HPA-5b Anti-HPA-5b Negative Negative Anti-HLA Anti-HPA-1a + anti-HLA Anti-HPA-1a + anti-HLA Anti-HPA-1b + anti-HLA Anti-HPA-1a
PAKLx first run Anti-GPIb/IX Anti-HLA + anti-GPIa/IIa + GPIIb/IIIa Anti-HPA-1a + anti-GP† Anti-HPA-1a + 5b + anti-GPIb/IX Anti-HPA-1a + 5b Anti-HPA-1a + 5b Anti-HPA-1a Anti-GPIa/IIa Anti-HPA-1a + 5b Anti-HPA-1a‡ + 5b Anti-HLA Anti-HLA Negative Anti-HPA-1a Anti-HPA-1a Anti-HPA-1b Anti-HPA-1a + anti-HLA
PAKLx repeat Anti-GPIb/IX Anti-HLA + anti-GPIa/IIa + GPIIb/IIIa Anti-HPA-1a + anti-GP† Anti-HPA-1a + anti-GPIb/IX Anti-HPA-1a Anti-HPA-1a Anti-HPA-1a Anti-GPIa/IIa Anti-HPA-5b Anti-HPA-1a‡ + 5b Anti-HLA Anti-HLA Negative Anti-HPA-1a Anti-HPA-1a Anti-HPA-1b Anti-HPA-1a + anti-HLA
* Bold text indicates false-negative or -positive anti-HPA or anti-GP results. Italic text indicates discrepant anti-HLA results. † All beads for GPIb/IX, Ia/IIa, and IIb/IIIa showed positive reactions. ‡ Already suspected anti-HPA-1a in serum of HPA-1bb5aa typed mother.
PAKLx (Table 4). Repeating these samples in the MAIPA showed the same results. After repeating these 10 samples twice in the PAKLx, two initial positive reactions with HPA-5b and one initial positive reaction with HPA-1a were not detected anymore (Table 4, Samples 5, 6, and 9). Therefore, we regard these initial positive reactions as false positive. Seven discrepancies remained. Four samples showed extra GP-reactive antibodies only in the PAKLx (Table 4, Samples 2-5), in one sample containing anti-HPA-1a plus -3a, the anti-HPA-3a was not detected in the PAKLx (Table 4, Sample 7), and for one sample the PAKLx showed broadly reactive GPIa/IIa antibodies instead of anti-HPA-5b (Table 4, Sample 8). One sample already suspected for the presence of anti-HPA-1a and -5b, because this woman was typed HPA-1bb5aa, showed negative results for the anti-HPA-1a in the MAIPA and in the PAKLx. In the PAKLx, however, the slightly increased MFI with homozygous HPA-1a beads suggested the presence of an anti-HPA-1a (Table 4, Sample 10). Seven samples were disconcordant in detection of HLA Class I antibodies. One sample showed HLA antibodies in the MAIPA but not in the PAKLx (Table 4, Sample 13). Three sera that contained anti-HPA-1 (two antiHPA-1a and one anti-HPA-1b) and anti-HLA tested negative for anti-HLA in the PAKLx (Table 4, Samples 14-16). Three samples (one with anti-HPA-1a) reacted negative in the MAIPA, whereas positive reactions with the HLA Class I–carrying beads were obtained in the PAKLx (Table 4, Samples 11, 12, and 17). Of the 74 serum samples from patients with refractoriness to PLT transfusions, 24 reacted negative and 32 reacted positive in both techniques (Table 5). In the 32 samples, the following antibodies were detected: antiHPA-1a plus anti-HLA (n = 1), anti-HPA-1b plus anti-HLA (n = 2), anti-HPA-2a plus anti-HLA (n = 1), anti-HPA-2b 4
TRANSFUSION Volume **, ** **
TABLE 5. Antibody detection in samples from patients with refractoriness for PLT transfusions (n = 74) PAKLx MAIPA Negative Positive
Negative 24 5
Positive 3 32 + 10*
* Ten discrepant positive results in MAIPA and PAKLx.
plus -5b plus anti-HLA (n = 1), anti-HPA-5a plus anti-HLA (n = 1), anti-HPA-5b with or without anti-HLA (n = 2), GP-reactive autoantibodies (n = 2), and anti-HLA (n = 22). Eighteen discrepant results were found, of which 11 were for GP- or HPA-reactive antibodies (Table 6). After retesting these samples in the MAIPA, GPIa/IIa antibodies in one sample were not detectable anymore, which was in concordance with the PAKLx results; the initial MAIPA result was regarded as false positive. A weakly reactive anti-HPA-2b (as was indicated by the PAKLx) was also found weakly reactive in the MAIPA with extinctions just below the cutoff value (Table 6, Samples 4 and 11). After retesting in the PAKLx, positive reactions with HPA-1a, HPA-3a, and GPIa/IIa were not detected anymore (Table 6, Samples 3, 5, and 8) and regarded as false positive. Seven discrepancies remained (Table 6, Samples 1, 2, 3, 6, 7, 9, and 10). Two MAIPA-negative samples showed GPIIb/IIIa-reactive antibodies in the PAKLx (Table 6, Samples 1 and 2), one anti-HPA-2b sample showed GPIa/ IIa antibodies in the PAKLx, one anti-HPA-2b plus -5b sample showed anti-HPA-2b plus GPIa/IIa in PAKLx, and one serum with HLA antibodies also showed the presence of anti-HPA-3a plus GPIIb/IIIa antibodies in the PAKLx (patient was typed HPA-3a positive; Table 6, Samples 6, 7, and 10). A weak anti-HPA-5b was not detected in the
BEAD-BASED HPA ANTIBODY DETECTION
TABLE 6. Discrepant results for samples of patients with refractoriness for PLT transfusions* Sample 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 * † ‡ § ||
MAIPA Negative Negative Negative† Anti-HPA-1b + 3a + 5a + GPIa/IIa‡ Anti-HPA-2b Anti-HPA-2b Anti-HPA-2b + 5b Anti-HPA-5a Anti-HPA-5b (OD 0.45) Anti-HLA Anti-HLA§ Anti-HLA Anti-HLA Anti-HLA Anti-HLA Anti-HPA-1a Anti-HPA-1a Anti-HLA + anti-GP||
PAKLx first run Anti-GPIIb/IIIa Anti-GPIIb/IIIa Anti-HPA-1a + GPIV Anti-HPA-1b + 3a + 5a Anti-HPA-2b + 3a Anti-HPA-2b + GPIa/IIa Anti-HPA-2b + GPIa/IIa Anti-HPA-5a + GPIa/IIa Negative Anti-HLA + anti-HPA-3a + GPIIb/IIIa Anti-HPA-2b§ + anti-HLA Negative Negative Negative Negative Anti-HPA-1a + anti-HLA Anti-HPA-1a + anti-HLA Anti-GP||
PAKLx repeat Anti-GPIIb/IIIa Anti-GPIIb/IIIa Anti-GPIV Anti-HPA-1b + 3a + 5a Anti-HPA-2b Anti-HPA-2b + GPIa/IIa Anti-HPA-2b + GPIa/IIa Anti-HPA-5a Negative Anti-HLA + anti-HPA-3a + GPIIb/IIIa Anti-HPA-2b§ + anti-HLA Negative Negative Negative Negative Anti-HPA-1a + anti-HLA Anti-HPA-1a + anti-HLA Anti-GP||
Bold text indicates false-negative or -positive anti-HPA or anti-GP results. Italic text indicates discrepant HLA antibody results. Anti-GPIV was detected in PIFT, but not in MAIPA. Repeated MAIPA showed no panreactive GPIa/IIa anymore. In both the repeated MAIPA and the repeated PAKLx the anti-HPA-2b was suspected. GPIb/IX, Ia/IIa, and IIb/IIIa showed positive reactions.
TABLE 7. Antibody detection in samples from healthy male donors (n = 20) PAKLx MAIPA Negative Positive
Negative 18 0
Positive 2* 0
* One anti-HPA-3a (only in the first run) and one anti-HLA.
PAKLx, which was regarded as a “false-negative” result (Table 6, Sample 9). An anti-GPIV (detected in the PIFT) was correctly detected with the PAKLx but missed by MAIPA (Table 6, Sample 3). Seven samples were discrepant for detection of HLA Class I antibodies. Five samples showed the presence of HLA antibodies only in the MAIPA but not in the PAKLx (Table 6, Samples 12-15 and 18); in two samples HLA antibodies were detected only by the PAKLx (next to HPA-1a antibodies; Table 6, Samples 16 and 17). For the three samples with PLT-reactive autoantibodies the results obtained by MAIPA were in full concordance with PAKLx. Serum samples from 20 healthy male blood donors were tested in the MAIPA and PAKLx. All samples were negative in the MAIPA. In the PAKLx one sample showed, only in the first run, an anti-HPA-3a and one sample showed anti-HLA (Table 7). The overall correlation (R2 = 0.67) between MAIPA ODs and PAKLx MFI results for anti-HPA-1a, -1b, -2a, -2b, -3a, -5a, and 5b antibodies is shown in Fig. 1A. As is shown in Figs. 1B and 1C, the correlation between MAIPA OD and PAKLx MFI results for anti-HPA-1a and -1b (R2 = 0.867) is better than for the HPA-5 antibodies (R2 = 0.594).
Testing 175 sera with the PAKLx showed, compared with the LMX, 171 (98%) concordant and four (2%) discrepant results. Sixty-five samples reacted positive and 106 samples reacted negative in both the PAKLx and the LMX. Three samples reacted positive in LMX and negative in PAKLx and one sample reacted negative in LMX and positive in PAKLx. Two PAKLx-negative and LMX-positive samples also showed positive reactions in the lymphocyte immunofluorescence technique (LIFT)14 and in the MAIPA, which suggests the presence of HLA antibodies. One PAKLx-negative and LMX-positive sample reacted negative in the LIFT (with a pooled donor lymphocyte suspension containing the most relevant Caucasian HLA Class I antigens) and MAIPA, pointing to the absence of HLA antibodies. The PAKLx-positive and LMX-negative sample reacted positive in the LIFT and MAIPA, indicating the presence HLA antibodies.
DISCUSSION Overall results showed four (6%) false-negative results in PAKLx in a total of 67 HPA or GP-reactive antibodies containing sera compared with the MAIPA assay. The PAKLx assay failed to detect an anti-HPA-3a (MAIPA OD 0.55) as this antibody was clouded by a strong reactive antiHPA-1a (MAIPA OD 2.8) in the same sample. It also failed to detect an anti-HPA-5b (MAIPA OD 0.45) and could not distinguish two anti-HPA-5b (MAIPA OD 0.55 and 0.7) but indicated instead GPIa/IIa specificity. Furthermore, a labile reactive anti-HPA-3a, which could only be detected by PIFT and by MAIPA using different MoAb was also not detected by PAKLx. Volume **, ** **
TRANSFUSION
5
PORCELIJN ET AL.
PAKLx MFI
A
HPA-specific antibodies 24000 22000 20000 18000 16000 14000 12000 10000 8000 6000 4000 2000 0
y = 6511.8x - 422.09 R2 = 0.6997
0
0.5
1
1.5
2
2.5
3
MAIPA OD
B
anti HPA-1a and -1b 25000
y = 7307.2x + 187.36 R2 = 0.867
PAKLx MFI
20000 15000 10000 5000 0 0
0.5
1
1.5
2
2.5
3
3.5
MAIPA OD
C
anti HPA-5a and -5b 20000 18000
y = 4946.1x - 271.51 R2 = 0.5935
PAKLx MFI
16000 14000 12000 10000 8000 6000 4000 2000 0 0
0.5
1
1.5
2
2.5
3
MAIPA OD
Fig. 1. MAIPA OD versus PAKLx MFI in detection of: (A) all tested HPA-specific antibodies, (B) anti-HPA-1a and anti-HPA1b, and (C) anti-HPA-5a and anti-HPA-5b.
PAKLx, compared to the MAIPA assay, showed eight (4.1%) false-positive GP reactions in a total of 194 sera, of which one also showed a false-positive anti-HPA-3a (patient was typed HPA-3a positive). The first analysis of the 194 samples in the PAKLx showed 18 positive reactions with GP and/or HPA specificity not detected by MAIPA. Six of these 18 positive samples showed negative results in the 6
TRANSFUSION Volume **, ** **
second and third analysis in the PAKLx. All runs were done by the same two technicians, incorrect results were seen in equal numbers for both technicians, and no test problems were noticed. To value the PAKLx intertest variability, more tests are needed. In total we obtained 12 of 194 (6.1%) incorrect anti-GP and/or anti-HPA results with the PAKLx. In one case the MAIPA did not show an anti-HPA-1a. This antiHPA-1a was also scored negative by the PAKLx software, but was suspected because of a slight difference in MFIs between the antigen-positive and -negative beads. The anti-HPA-1a was already suspected in an FNAIT case as the mother was HPA-1a negative. The MAIPA also did not pick up a strong reactive (in the PIFT) anti-GPIV, possibly due to inhibition of binding of CD36 MoAb. The antiGPIV showed a strong reaction with the GPIV-coated beads in the PAKLx. Differences in detection of HLA antibodies was noted in 13 samples, but since the MAIPA is not meant for HLA antibody screening, the HLA bead reactivity in the PAKLx setting was compared with the reactivity of the comparable bead in the LMX. A 98% concordance between the PAKLx and “bead one” of LMX was seen. The adjusted ratio values in the PAKLx with which the software calls positive or negative are indicative, but careful observation of the results is necessary to detect weak antibodies. For instance, weak antibody reactions only positive with homozygous beads but not with heterozygous beads are not picked up by the software and can only be detected upon manual judgment of the signals. The higher background level obtained with the antiHPA-5 detection beads impairs the detection and exclusion of the presence of weakly reactive antibodies. The PAKLx test takes approximately 3 hours to perform, does not require typed donor PLTs or MoAb, and only uses 10 μL of serum for antibody screening plus identification. These are major advantages compared to the MAIPA, which takes 6 to 8 hours and is quite difficult to perform, needing skilled technicians, a typed donor PLT identification panel, GP-specific MoAbs, and higher amounts of patient serum. At this moment the PAKLx does not include HPA-15 antibody identification beads and confirmation of the antibody specificity with the available beads is not always possible. For HPA-5 antibodies, weak antibody reactions are sometimes difficult to distinguish from background MFI values, which might be the result of the lower correlation between the MAIPA OD and PAKLx MFI. In our study we focused on the detection of PLTreactive alloantibodies. It will be interesting to assess the wider utility of this assay for the detection of GP-reactive autoantibodies and isoantibodies. In our panel of sera the very rare HPA-3b specificity was not included and some other HPA specificities (HPA-2a, -3a, and -5a and GPIV) were only present in a limited number of samples. Therefore, it will be necessary to further evaluate this assay
BEAD-BASED HPA ANTIBODY DETECTION
with an extended number of samples with (rare) HPA specificities. We conclude that the PAKLx is an easy-to-perform sensitive PLT alloantibody screening method. In case of weakly reactive antibodies or detection of broadly reactive antibodies confirmation of the antibody specificity with the MAIPA should be performed, because aspecific reactions were detected in the PAKLx.
(MAIPA). A new tool for the identification of plateletreactive antibodies. Blood 1987;70:1722-6. 6. Kaye T, Allen DL. Identification of platelet-reactive antibodies using an antigen-capture ELISA. Transfus Med 1994;4(Suppl 1):49. 7. Aster RH. Evaluation of GTI-PAK1 antigen capture ELISA (ACE) for the detection of IgG antibodies to plateletspecific antigens. Platelets 1994;5:222. 8. Porcelijn L, van Beers W, Gratama JW, et al. External quality assessment of platelet serology and human platelet
AUTHORSHIP LP was the principal investigator and author; EH was the chief technician; IL and AC were technicians; VP was the supervisor at Immucor-GTI Diagnostics; MM was the technician at HOLOGIC; and MdH was the supervisor at Sanquin Diagnostic Services.
antigen genotyping: a 10-year review. Transfusion 2008;48: 1699-706. 9. Bakchoul T, Meyer O, Agaylan A, et al. Rapid detection of HPA-1 alloantibodies by platelet antigens immobilized onto microbeads. Transfusion 2007;47:1363-8. 10. Chong W, Metcalfe P, Mushens R, et al. Detection of
CONFLICT OF INTEREST
human platelet antigen-1a alloantibodies in cases of
VR and MM are employees of Immucor-GTI Diagnostics and
fetomaternal alloimmune thrombocytopenia using recombinant β3 integrin fragments coupled to fluorescently labeled beads. Transfusion 2011;51:1261-70.
HOLOGIC, Waukesha, WI. The other authors have no conflict of interest.
REFERENCES 1. Serrarens-Janssen VM, Semmekrot BA, Novotny VM, et al. Fetal/neonatal allo-immune thrombocytopenia (FNAIT): past, present, and future. Obstet Gynecol Surv 2008;63:23952. 2. Hod E, Schwartz J. Platelet transfusion refractoriness. Br J Haematol 2008;142:348-60. 3. Shtalrid M, Shvidel L, Vorst E, et al. Post-transfusion purpura: a challenging diagnosis. Isr Med Assoc J 2006;8: 672-4. 4. von dem Borne AE, Verheugt FW, Oosterhof F, et al. A simple immunofluorescence test for the detection of platelet antibodies. Br J Haematol 1978;39:195-207. 5. Kiefel V, Santoso S, Weisheit M, et al. Monoclonal antibody-specific immobilization of platelet antigens
11. Skaik Y, Battermann A, Hiller O, et al. Development of a single-antigen magnetic bead assay (SAMBA) for the sensitive detection of HPA-1a alloantibodies using tagengineered recombinant soluble β3 integrin. J Immunol Methods 2013;391(1-2):72-80. 12. Fujiwara K, Shimano K, Tanaka H, et al. Application of bead array technology to simultaneous detection of human leucocyte antigen and human platelet antigen antibodies. Vox Sang 2009;96:244-51. 13. Colombo MB, Haworth SE, Poli F, et al. Luminex technology for anti-HLA antibody screening: evaluation of performance and of impact on laboratory routine. Cytometry B Clin Cytom 2007;72:465-71. 14. Décary F, Vermeulen A, Engelfriet CP. A look at HLA antisera in the indirect immunofluorescence technique (IIFT). In: Kissmeyer-Nielsen F, editor. Histocompatibility testing. Copenhagen: Munksgaard; 1975. p. 380-90.
Volume **, ** **
TRANSFUSION
7
BACKGROUND: The performance of a newly developed Luminex bead-based platelet (PLT) antibody detection method (PAKLx) was compared with the monoclonal antibody immobilization of PLT antigens (MAIPA) assay and the LifeScreen Deluxe Luminex bead-based HLA Class I antibody detection method (LMX). STUDY DESIGN AND METHODS: Six sera containing anti-human PLT antigen (HPA)-1a (n = 2), HPA-1b, HPA-2b, HPA-3a, or HPA-5b were tested in titration. A total of 194 sera, including HPA-1a, -1b, -2a, -2b, -3a, -5a, and -5b antibodies with or without HLA antibodies (n = 63); glycoprotein (GP) IV antibodies (n = 1); PLT autoantibodies (n = 3); HLA antibodies (n = 45); and samples with no PLT-reactive antibodies (n = 82), were tested in both assays. RESULTS: Comparable levels of sensitivity were obtained for the MAIPA and PAKLx. The PAKLx showed four (6%) false-negative results in 67 sera with HPA or GP-reactive antibodies: anti-HPA-3a (n = 1) or antiHPA-5b (n = 3). The PAKLx showed in 10 of the total 194 samples (5%) the presence of antibodies not detected by the MAIPA. This concerned broadly GP-reactive antibodies (n = 7), anti-GPIIb/IIIa combined with anti-HPA-3a (n = 1), anti-HPA-1a (borderline, n = 1), and anti-GPIV (n = 1). Testing 175 sera for antiHLA Class I antibodies in the PAKLx and LMX showed four discrepant results: PAKLx negative and LMX positive, n = 3 and n = 1, respectively. CONCLUSION: For the vast majority of the specimens tested (93%) the results of the PAKLx were in concordance with the MAIPA. The PAKLx is a fast, easy to perform, and sensitive PLT antibody screening method.
A
lloantibodies directed against human platelet antigens (HPA) are of clinical significance in fetal and neonatal alloimmune thrombocytopenia (FNAIT),1 refractoriness to platelet (PLT) transfusions,2 and posttransfusion purpura.3 Currently, several techniques are used for the detection of PLT-specific alloantibodies, such as the PLT immunofluorescence test (PIFT),4 the monoclonal antibody immobilization of PLT antigens (MAIPA) assay,5 and various types of solid-phase enzyme-linked immunosorbent assay using PLT-specific glycoproteins (GPs).6,7 Until now the MAIPA is the most sensitive and specific PLT alloantibody detection technique.8 The MAIPA is a laborious assay that takes 6 to 8 hours and due to the necessity to use HPA-typed PLTs; it can only be performed in specialized reference laboratories.
ABBREVIATIONS: FNAIT = fetal and neonatal alloimmune thrombocytopenia; GP = glycoprotein; LIFT = lymphocyte immunofluorescence technique; LMX = LifeScreen Deluxe Luminex bead-based HLA Class I antibody detection method; MAIPA = monoclonal antibody immobilization of platelet antigens; PIFT = platelet immunofluorescence test. From the 1Sanquin Diagnostic Services, Amsterdam, The Netherlands; and 2Immucor-GTI Diagnostics and 3HOLOGIC, Waukesha, Wisconsin. Address reprint requests to: Leendert Porcelijn, MD, Department of Immunohematology Diagnostic Services, Platelet/Leukocyte Serology Laboratory, Sanquin Diagnostic Services, Plesmanlaan 125, 1066CX Amsterdam, The Netherlands; e-mail: [email protected]. The PAKLx tests were provided by Gen-Probe Diagnostics, Inc. and PAKLx materials were provided by Lifecodes products, HOLOGIC/Gen-Probe. Received for publication July 11, 2013; revision received October 4, 2013, and accepted October 22, 2013. doi: 10.1111/trf.12509 © 2013 AABB TRANSFUSION **;**:**-**. Volume **, ** **
TRANSFUSION
1
PORCELIJN ET AL.
Bead-based HPA antibody detection assays for only the detection of anti-HPA-1 have been described by three different groups. Bakchoul and colleagues in 20079 used purified GPIIb/IIIa from outdated PLTs in a gel antigenspecific assay, Chong and colleagues in 201010 used recombinant β3-integrins displaying the HPA-1a or HPA-1b epitope, and Skaik and colleagues in 201311 used recombinant soluble β3-integrins displaying HPA-1a or HPA-1b epitopes in a single-antigen magnetic bead assay. Fujiwara and coworkers described in 200912 the immunecomplex capture fluorescence analysis, a kind of beadbased MAIPA in which they used microarray beads coupled with monoclonal antibodies (MoAbs) specific for CD36, CD41, CD42b, CD49b, CD61, and HLA Class I antigens to capture GP-antibody complexes. Recently, the PAKLx bead-based GP-specific HPA antibody detection method (Lifecodes products, HOLOGIC/Gen-Probe, Waukesha, WI) became available. This assay uses beads to which PLT lysate–derived GPs carrying clinical relevant HPA or HLA Class I specificities are coupled to capture and identify HPA and HLA Class I antibodies. This assay is easy to perform in 3 hours and uses only 10 μL of serum. We compared MAIPA and PAKLx test results for 194 serum samples from women who gave birth to a child suspected to suffer from FNAIT (n = 100), patients with refractoriness for PLT transfusions (n = 74), and healthy male donors (n = 20).
MATERIALS AND METHODS Patient samples Six sera with HPA antibodies of different specificities (see Table 1) were tested in titration in the MAIPA and
Antibody detection assays
TABLE 1. PAKLx test beads Sample 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
2
GP group Control bead 1 Control bead 2 Control bead 3 Positive control bead GPIV HLA Class I GPIIb-IIIa GPIIb-IIIa GPIIb-IIIa GPIIb-IIIa GPIIb-IIIa GPIIb-IIIa GPIb-IX GPIb-IX GPIb-IX GPIb-IX GPIb-IX GPIa-IIa GPIa-IIa GPIa-IIa GPIa-IIa GPIa-IIa
TRANSFUSION Volume **, ** **
PAKLx. A total of 100 sera from women who gave birth to a child suspected to suffer from FNAIT, 74 sera from patients who showed PLT refractoriness to PLT transfusions (refractoriness cases), and 20 sera from healthy male (never transfused) donors were tested in the MAIPA and PAKLx. Serum samples from 174 cases were referred to the laboratory of Platelet and Leukocyte Serology of Sanquin Diagnostic Services for FNAIT or refractoriness to PLT transfusion investigation. One or more anti-HPA and/or GP-reactive antibodies (with or without HLA antibodies) were present in 67 of the 174 samples (as detected in the MAIPA); anti-HPA-1a (n = 29; optical density [OD] in MAIPA varying from 0.5 to >3.0 [cut-off level OD, 0.15]), anti-HPA-1b (n = 5; OD in MAIPA varying from 0.15 to >3.0), anti-HPA-2a (n = 1; OD, 2.5), anti-HPA-2b (n = 4; OD varying from 0.25 to 1.2), antiHPA-3a (n = 4; OD varying from 0.4 to >3.0 for three of these antibodies and one labile anti-HPA-3a only detectable in the PIFT as described by von dem Borne et al.4 and in the MAIPA if using some other, not routinely used, MoAbs, see below), anti-HPA-5a (n = 6; OD varying from 0.4 to 2.3), anti-HPA-5b (n = 16; OD varying from 0.2 to 2.5), autoantibodies (n = 3; two samples scored as weak positive in MAIPA [OD, 0.2-0.4] with CD49b/CD29 [GPIa/ IIa], CD42 [GPIb/IX], and CD41/CD61 [GPIIb/IIIa] and one sample scored as weak positive in MAIPA [OD, 0.2] with GPIa/IIa and GPIb/IX), and anti-GPIV (CD36, n = 1; not detectable in the MAIPA, but reactive in the PIFT with a panel of typed donor PLT suspensions). Of the 174 collected samples, according to the regulation of Sanquin Diagnostic Services, anonymized leftover materials were used. Written informed consent was obtained for the 20 donor samples.
Antigens
GPIV HLA Class I HPA-1a-3a-4a HPA-1a-3b-4a HPA-1b-3a-4a HPA-1b-3b-4a HPA-1ab-3ab-4a HPA-1a-3ab-4b HPA-2a HPA-2a HPA-2ab HPA-2b HPA-2b HPA-5a HPA-5a HPA-5ab HPA-5b HPA-5b
The PAKLx (Lifecodes products, HOLOGIC/Gen-Probe) is a bead-based technique for detection and specification of antibodies against antigens located on GPIIb/IIIa—HPA1a, -1b, -3a, -3b, -4a, and -4b; on GPIb/IX—2a and 2b; on GPIa/IIa—5a and 5b; GPIV (beads coated with affinity purified GPs from HPA typed PLT donors); and HLA Class I (bead coated with affinity purified HLA Class I obtained from a PLT pool that was composed of 100 Caucasian, 100 African American, and 100 Hispanic blood donors; Table 1). A kit is supplied containing 800 μL of beads. The beads include a positive control, three negative control beads, and 18 HPA or HLA Class I–specific beads, 100 μL of phycoerythrin (PE)-conjugated goat anti-human immunoglobulin (Ig)G, 1 mL of diluent, 30 mL of phosphate-based wash buffer, 80 μL of positive control serum (anti-HPA-1a), and 80 μL of negative control serum (inert serum). Forty microliters of pooled beads is incubated with 10 μL of patient serum for 60 minutes and washed four times with wash buffer, after which 50 μL of
BEAD-BASED HPA ANTIBODY DETECTION
PE-labeled anti-human IgG is added. After 30 minutes of incubation, antibody binding is measured using a Luminex instrument. Specific software is provided to determine if a result is positive. The mean fluorescence intensity (MFI) of the individual bead is divided by the MFI for each negative control, after which a background adjustment factor is subtracted for the appropriate bead/ control bead combination (bead MFI/negative control bead 1 MFI minus background adjustment factor = adjusted ratio 1, idem for 2 and 3). If two or three of the adjusted ratios are positive, the result is positive. The MAIPA used is a modification of the MAIPA described by Kiefel and colleagues.5 In short, a panel of typed donor PLTs (15 × 106 if testing for GPIIb/IIIa and 40 × 106 for the other GP) were incubated with 120 μL of patient serum (maternal sera for FNAIT cases) and a MoAb (all from our institution, Division Reagents of Sanquin Blood Supply) recognizing GPIIb/IIIa (CLBthromb/1, CD41, anti-GPIIb), GPIb/IX (MB45, CD42a, anti-GPIX), GPIa/IIa (10G11, CD49b, anti-GPIa), GPIV (P58, CD36), or HLA Class I (W6.32). For the detection of one labile anti-HPA-3a (see below) we also used the GPIIb/IIIa-specific MoAb Y2/51 and PL164-A101 (CD61 and CD41, MoAb workshop Vienna, 1989) and CLBthromb/7 (CD41, Division Reagents of Sanquin Blood Supply). After binding of the MoAb, the PLTs were solubilized with an Nonidet P-40–containing buffer. The lysates were added to a microtiter plate coated with goat immunoglobulin anti-mouse IgG to immobilize the MoAb-GP-human antibody complexes. Antibodies bound to the analyzed GP or GP-complex are detected with horseradish peroxidase labeled goat immunoglobulin anti-human IgG. The samples were tested for antibodies against HPA-1a, -1b, and -3a (GPIIb/IIIa); HPA-2a and 2b (GPIb/IX); HPA-5a and -5b (GPIa/IIa); and HLA Class I and GPIV. Discrepant results were retested twice in the PAKLx to finally use the results detected in two of three tests, and once in the MAIPA to confirm the initial MAIPA results to exclude sample mix-up or loss or gain of signals. Because the MAIPA is not routinely used for the detection of Class I HLA-reactive antibodies, the LifeScreen Deluxe (LMX) assay, a bead-based immunoassay, was used for the detection of HLA-reactive antibodies.13 All 194 samples used in the comparison were first qualified for the study by performance in LMX. Any sample that produced indeterminate, borderline, and/or ambiguous results in LMX were excluded from the study for consideration, leaving 175 LMX well-defined samples. These 175 samples were tested both in the PAKLx with the HLA Class I–carrying bead (coated with affinitypurified HLA Class I GPs obtained from PLTs of 100 Caucasian, 100 African American, and 100 Hispanic blood donors) and with the comparable HLA Class I–coated bead in the LMX.
Statistical analysis Correlations between MAIPA ODs and PAKLx MFIs were calculated with computer software (Microsoft Excel 2002, Microsoft Corp., Redmond, WA).
RESULTS Anti-HPA-1a (n = 2), anti-HPA-1b, anti-HPA-3a, and antiHPA-5a were all correctly detectable in the first test of six selected sera with the PAKLx. Tested in dilution, sera containing anti-HPA-1a (n = 2), anti-HPA-1b (n = 1), anti-HPA-3a (n = 1), and anti-HPA-5a (n = 1) all showed comparable results in the MAIPA and PAKLx (Table 2). Anti-HPA-2b was detectable up to a dilution of 32 in the MAIPA compared to a dilution of 8 in the PAKLx. Of the 100 maternal sera from FNAIT cases, 39 showed negative results both in the MAIPA and in the PAKLx (Table 3). In one of these 39 samples an antiHPA-3a was detected in the PIFT. This HPA-3a antibody was not detected by PAKLx or by our routine MAIPA using the MoAb CLBThromb/1. However, it did react in the MAIPA using the MoAb Y2, 6C9, and P21 (OD varying from 0.16 to 0.90). Forty-four sera reacted positive in both techniques. In these sera the following antibody specificities were found: anti-HPA-1a with or without anti-HLA (n = 17), antiHPA-1b with anti-HLA (n = 1), anti-HPA-5a with or without anti-HLA (n = 4), anti-HPA-5b with or without anti-HLA (n = 10), anti-HPA-1a with -5b with anti-HLA (n = 1), anti-HPA-3a with -5b with anti-HLA (n = 1), and anti-HLA (n = 10). The 17 samples showing discrepant results between MAIPA and PAKLx are shown in Table 4. Ten samples showed results discrepant for the detection of GP or HPA-reactive antibodies between MAIPA and
TABLE 2. Six sera containing antibodies with a HPA specificity tested in titration MAIPA results HPA-1a HPA-1a HPA-1b HPA-2b HPA-3a HPA-5b
MAIPA titer
PAKLx results
PAKLx titer
>128* 256 32 32 1024 1024
Anti-HPA-1a Anti-HPA-1a Anti-HPA-1b Anti-HPA-2b Anti-HPA-3a Anti-HPA-5b
>128* 256 32 8 1024 1024
* The serum was not further diluted.
TABLE 3. Antibody detection in samples from possible FNAIT cases (n = 100) PAKLx MAIPA Negative Positive
Negative
Positive
39 1
3 44 + 13*
* Thirteen discrepant positive results in MAIPA and PAKLx.
Volume **, ** **
TRANSFUSION
3
PORCELIJN ET AL.
TABLE 4. Discrepant results obtained with sera from cases analyzed because of possible FNAIT* Sample 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
MAIPA Negative Anti-HLA Anti-HPA-1a Anti-HPA-1a Anti-HPA-1a Anti-HPA-1a Anti-HPA-1a + 3a (OD 0.55) Anti-HPA-5b (OD 0.55) Anti-HPA-5b Anti-HPA-5b Negative Negative Anti-HLA Anti-HPA-1a + anti-HLA Anti-HPA-1a + anti-HLA Anti-HPA-1b + anti-HLA Anti-HPA-1a
PAKLx first run Anti-GPIb/IX Anti-HLA + anti-GPIa/IIa + GPIIb/IIIa Anti-HPA-1a + anti-GP† Anti-HPA-1a + 5b + anti-GPIb/IX Anti-HPA-1a + 5b Anti-HPA-1a + 5b Anti-HPA-1a Anti-GPIa/IIa Anti-HPA-1a + 5b Anti-HPA-1a‡ + 5b Anti-HLA Anti-HLA Negative Anti-HPA-1a Anti-HPA-1a Anti-HPA-1b Anti-HPA-1a + anti-HLA
PAKLx repeat Anti-GPIb/IX Anti-HLA + anti-GPIa/IIa + GPIIb/IIIa Anti-HPA-1a + anti-GP† Anti-HPA-1a + anti-GPIb/IX Anti-HPA-1a Anti-HPA-1a Anti-HPA-1a Anti-GPIa/IIa Anti-HPA-5b Anti-HPA-1a‡ + 5b Anti-HLA Anti-HLA Negative Anti-HPA-1a Anti-HPA-1a Anti-HPA-1b Anti-HPA-1a + anti-HLA
* Bold text indicates false-negative or -positive anti-HPA or anti-GP results. Italic text indicates discrepant anti-HLA results. † All beads for GPIb/IX, Ia/IIa, and IIb/IIIa showed positive reactions. ‡ Already suspected anti-HPA-1a in serum of HPA-1bb5aa typed mother.
PAKLx (Table 4). Repeating these samples in the MAIPA showed the same results. After repeating these 10 samples twice in the PAKLx, two initial positive reactions with HPA-5b and one initial positive reaction with HPA-1a were not detected anymore (Table 4, Samples 5, 6, and 9). Therefore, we regard these initial positive reactions as false positive. Seven discrepancies remained. Four samples showed extra GP-reactive antibodies only in the PAKLx (Table 4, Samples 2-5), in one sample containing anti-HPA-1a plus -3a, the anti-HPA-3a was not detected in the PAKLx (Table 4, Sample 7), and for one sample the PAKLx showed broadly reactive GPIa/IIa antibodies instead of anti-HPA-5b (Table 4, Sample 8). One sample already suspected for the presence of anti-HPA-1a and -5b, because this woman was typed HPA-1bb5aa, showed negative results for the anti-HPA-1a in the MAIPA and in the PAKLx. In the PAKLx, however, the slightly increased MFI with homozygous HPA-1a beads suggested the presence of an anti-HPA-1a (Table 4, Sample 10). Seven samples were disconcordant in detection of HLA Class I antibodies. One sample showed HLA antibodies in the MAIPA but not in the PAKLx (Table 4, Sample 13). Three sera that contained anti-HPA-1 (two antiHPA-1a and one anti-HPA-1b) and anti-HLA tested negative for anti-HLA in the PAKLx (Table 4, Samples 14-16). Three samples (one with anti-HPA-1a) reacted negative in the MAIPA, whereas positive reactions with the HLA Class I–carrying beads were obtained in the PAKLx (Table 4, Samples 11, 12, and 17). Of the 74 serum samples from patients with refractoriness to PLT transfusions, 24 reacted negative and 32 reacted positive in both techniques (Table 5). In the 32 samples, the following antibodies were detected: antiHPA-1a plus anti-HLA (n = 1), anti-HPA-1b plus anti-HLA (n = 2), anti-HPA-2a plus anti-HLA (n = 1), anti-HPA-2b 4
TRANSFUSION Volume **, ** **
TABLE 5. Antibody detection in samples from patients with refractoriness for PLT transfusions (n = 74) PAKLx MAIPA Negative Positive
Negative 24 5
Positive 3 32 + 10*
* Ten discrepant positive results in MAIPA and PAKLx.
plus -5b plus anti-HLA (n = 1), anti-HPA-5a plus anti-HLA (n = 1), anti-HPA-5b with or without anti-HLA (n = 2), GP-reactive autoantibodies (n = 2), and anti-HLA (n = 22). Eighteen discrepant results were found, of which 11 were for GP- or HPA-reactive antibodies (Table 6). After retesting these samples in the MAIPA, GPIa/IIa antibodies in one sample were not detectable anymore, which was in concordance with the PAKLx results; the initial MAIPA result was regarded as false positive. A weakly reactive anti-HPA-2b (as was indicated by the PAKLx) was also found weakly reactive in the MAIPA with extinctions just below the cutoff value (Table 6, Samples 4 and 11). After retesting in the PAKLx, positive reactions with HPA-1a, HPA-3a, and GPIa/IIa were not detected anymore (Table 6, Samples 3, 5, and 8) and regarded as false positive. Seven discrepancies remained (Table 6, Samples 1, 2, 3, 6, 7, 9, and 10). Two MAIPA-negative samples showed GPIIb/IIIa-reactive antibodies in the PAKLx (Table 6, Samples 1 and 2), one anti-HPA-2b sample showed GPIa/ IIa antibodies in the PAKLx, one anti-HPA-2b plus -5b sample showed anti-HPA-2b plus GPIa/IIa in PAKLx, and one serum with HLA antibodies also showed the presence of anti-HPA-3a plus GPIIb/IIIa antibodies in the PAKLx (patient was typed HPA-3a positive; Table 6, Samples 6, 7, and 10). A weak anti-HPA-5b was not detected in the
BEAD-BASED HPA ANTIBODY DETECTION
TABLE 6. Discrepant results for samples of patients with refractoriness for PLT transfusions* Sample 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 * † ‡ § ||
MAIPA Negative Negative Negative† Anti-HPA-1b + 3a + 5a + GPIa/IIa‡ Anti-HPA-2b Anti-HPA-2b Anti-HPA-2b + 5b Anti-HPA-5a Anti-HPA-5b (OD 0.45) Anti-HLA Anti-HLA§ Anti-HLA Anti-HLA Anti-HLA Anti-HLA Anti-HPA-1a Anti-HPA-1a Anti-HLA + anti-GP||
PAKLx first run Anti-GPIIb/IIIa Anti-GPIIb/IIIa Anti-HPA-1a + GPIV Anti-HPA-1b + 3a + 5a Anti-HPA-2b + 3a Anti-HPA-2b + GPIa/IIa Anti-HPA-2b + GPIa/IIa Anti-HPA-5a + GPIa/IIa Negative Anti-HLA + anti-HPA-3a + GPIIb/IIIa Anti-HPA-2b§ + anti-HLA Negative Negative Negative Negative Anti-HPA-1a + anti-HLA Anti-HPA-1a + anti-HLA Anti-GP||
PAKLx repeat Anti-GPIIb/IIIa Anti-GPIIb/IIIa Anti-GPIV Anti-HPA-1b + 3a + 5a Anti-HPA-2b Anti-HPA-2b + GPIa/IIa Anti-HPA-2b + GPIa/IIa Anti-HPA-5a Negative Anti-HLA + anti-HPA-3a + GPIIb/IIIa Anti-HPA-2b§ + anti-HLA Negative Negative Negative Negative Anti-HPA-1a + anti-HLA Anti-HPA-1a + anti-HLA Anti-GP||
Bold text indicates false-negative or -positive anti-HPA or anti-GP results. Italic text indicates discrepant HLA antibody results. Anti-GPIV was detected in PIFT, but not in MAIPA. Repeated MAIPA showed no panreactive GPIa/IIa anymore. In both the repeated MAIPA and the repeated PAKLx the anti-HPA-2b was suspected. GPIb/IX, Ia/IIa, and IIb/IIIa showed positive reactions.
TABLE 7. Antibody detection in samples from healthy male donors (n = 20) PAKLx MAIPA Negative Positive
Negative 18 0
Positive 2* 0
* One anti-HPA-3a (only in the first run) and one anti-HLA.
PAKLx, which was regarded as a “false-negative” result (Table 6, Sample 9). An anti-GPIV (detected in the PIFT) was correctly detected with the PAKLx but missed by MAIPA (Table 6, Sample 3). Seven samples were discrepant for detection of HLA Class I antibodies. Five samples showed the presence of HLA antibodies only in the MAIPA but not in the PAKLx (Table 6, Samples 12-15 and 18); in two samples HLA antibodies were detected only by the PAKLx (next to HPA-1a antibodies; Table 6, Samples 16 and 17). For the three samples with PLT-reactive autoantibodies the results obtained by MAIPA were in full concordance with PAKLx. Serum samples from 20 healthy male blood donors were tested in the MAIPA and PAKLx. All samples were negative in the MAIPA. In the PAKLx one sample showed, only in the first run, an anti-HPA-3a and one sample showed anti-HLA (Table 7). The overall correlation (R2 = 0.67) between MAIPA ODs and PAKLx MFI results for anti-HPA-1a, -1b, -2a, -2b, -3a, -5a, and 5b antibodies is shown in Fig. 1A. As is shown in Figs. 1B and 1C, the correlation between MAIPA OD and PAKLx MFI results for anti-HPA-1a and -1b (R2 = 0.867) is better than for the HPA-5 antibodies (R2 = 0.594).
Testing 175 sera with the PAKLx showed, compared with the LMX, 171 (98%) concordant and four (2%) discrepant results. Sixty-five samples reacted positive and 106 samples reacted negative in both the PAKLx and the LMX. Three samples reacted positive in LMX and negative in PAKLx and one sample reacted negative in LMX and positive in PAKLx. Two PAKLx-negative and LMX-positive samples also showed positive reactions in the lymphocyte immunofluorescence technique (LIFT)14 and in the MAIPA, which suggests the presence of HLA antibodies. One PAKLx-negative and LMX-positive sample reacted negative in the LIFT (with a pooled donor lymphocyte suspension containing the most relevant Caucasian HLA Class I antigens) and MAIPA, pointing to the absence of HLA antibodies. The PAKLx-positive and LMX-negative sample reacted positive in the LIFT and MAIPA, indicating the presence HLA antibodies.
DISCUSSION Overall results showed four (6%) false-negative results in PAKLx in a total of 67 HPA or GP-reactive antibodies containing sera compared with the MAIPA assay. The PAKLx assay failed to detect an anti-HPA-3a (MAIPA OD 0.55) as this antibody was clouded by a strong reactive antiHPA-1a (MAIPA OD 2.8) in the same sample. It also failed to detect an anti-HPA-5b (MAIPA OD 0.45) and could not distinguish two anti-HPA-5b (MAIPA OD 0.55 and 0.7) but indicated instead GPIa/IIa specificity. Furthermore, a labile reactive anti-HPA-3a, which could only be detected by PIFT and by MAIPA using different MoAb was also not detected by PAKLx. Volume **, ** **
TRANSFUSION
5
PORCELIJN ET AL.
PAKLx MFI
A
HPA-specific antibodies 24000 22000 20000 18000 16000 14000 12000 10000 8000 6000 4000 2000 0
y = 6511.8x - 422.09 R2 = 0.6997
0
0.5
1
1.5
2
2.5
3
MAIPA OD
B
anti HPA-1a and -1b 25000
y = 7307.2x + 187.36 R2 = 0.867
PAKLx MFI
20000 15000 10000 5000 0 0
0.5
1
1.5
2
2.5
3
3.5
MAIPA OD
C
anti HPA-5a and -5b 20000 18000
y = 4946.1x - 271.51 R2 = 0.5935
PAKLx MFI
16000 14000 12000 10000 8000 6000 4000 2000 0 0
0.5
1
1.5
2
2.5
3
MAIPA OD
Fig. 1. MAIPA OD versus PAKLx MFI in detection of: (A) all tested HPA-specific antibodies, (B) anti-HPA-1a and anti-HPA1b, and (C) anti-HPA-5a and anti-HPA-5b.
PAKLx, compared to the MAIPA assay, showed eight (4.1%) false-positive GP reactions in a total of 194 sera, of which one also showed a false-positive anti-HPA-3a (patient was typed HPA-3a positive). The first analysis of the 194 samples in the PAKLx showed 18 positive reactions with GP and/or HPA specificity not detected by MAIPA. Six of these 18 positive samples showed negative results in the 6
TRANSFUSION Volume **, ** **
second and third analysis in the PAKLx. All runs were done by the same two technicians, incorrect results were seen in equal numbers for both technicians, and no test problems were noticed. To value the PAKLx intertest variability, more tests are needed. In total we obtained 12 of 194 (6.1%) incorrect anti-GP and/or anti-HPA results with the PAKLx. In one case the MAIPA did not show an anti-HPA-1a. This antiHPA-1a was also scored negative by the PAKLx software, but was suspected because of a slight difference in MFIs between the antigen-positive and -negative beads. The anti-HPA-1a was already suspected in an FNAIT case as the mother was HPA-1a negative. The MAIPA also did not pick up a strong reactive (in the PIFT) anti-GPIV, possibly due to inhibition of binding of CD36 MoAb. The antiGPIV showed a strong reaction with the GPIV-coated beads in the PAKLx. Differences in detection of HLA antibodies was noted in 13 samples, but since the MAIPA is not meant for HLA antibody screening, the HLA bead reactivity in the PAKLx setting was compared with the reactivity of the comparable bead in the LMX. A 98% concordance between the PAKLx and “bead one” of LMX was seen. The adjusted ratio values in the PAKLx with which the software calls positive or negative are indicative, but careful observation of the results is necessary to detect weak antibodies. For instance, weak antibody reactions only positive with homozygous beads but not with heterozygous beads are not picked up by the software and can only be detected upon manual judgment of the signals. The higher background level obtained with the antiHPA-5 detection beads impairs the detection and exclusion of the presence of weakly reactive antibodies. The PAKLx test takes approximately 3 hours to perform, does not require typed donor PLTs or MoAb, and only uses 10 μL of serum for antibody screening plus identification. These are major advantages compared to the MAIPA, which takes 6 to 8 hours and is quite difficult to perform, needing skilled technicians, a typed donor PLT identification panel, GP-specific MoAbs, and higher amounts of patient serum. At this moment the PAKLx does not include HPA-15 antibody identification beads and confirmation of the antibody specificity with the available beads is not always possible. For HPA-5 antibodies, weak antibody reactions are sometimes difficult to distinguish from background MFI values, which might be the result of the lower correlation between the MAIPA OD and PAKLx MFI. In our study we focused on the detection of PLTreactive alloantibodies. It will be interesting to assess the wider utility of this assay for the detection of GP-reactive autoantibodies and isoantibodies. In our panel of sera the very rare HPA-3b specificity was not included and some other HPA specificities (HPA-2a, -3a, and -5a and GPIV) were only present in a limited number of samples. Therefore, it will be necessary to further evaluate this assay
BEAD-BASED HPA ANTIBODY DETECTION
with an extended number of samples with (rare) HPA specificities. We conclude that the PAKLx is an easy-to-perform sensitive PLT alloantibody screening method. In case of weakly reactive antibodies or detection of broadly reactive antibodies confirmation of the antibody specificity with the MAIPA should be performed, because aspecific reactions were detected in the PAKLx.
(MAIPA). A new tool for the identification of plateletreactive antibodies. Blood 1987;70:1722-6. 6. Kaye T, Allen DL. Identification of platelet-reactive antibodies using an antigen-capture ELISA. Transfus Med 1994;4(Suppl 1):49. 7. Aster RH. Evaluation of GTI-PAK1 antigen capture ELISA (ACE) for the detection of IgG antibodies to plateletspecific antigens. Platelets 1994;5:222. 8. Porcelijn L, van Beers W, Gratama JW, et al. External quality assessment of platelet serology and human platelet
AUTHORSHIP LP was the principal investigator and author; EH was the chief technician; IL and AC were technicians; VP was the supervisor at Immucor-GTI Diagnostics; MM was the technician at HOLOGIC; and MdH was the supervisor at Sanquin Diagnostic Services.
antigen genotyping: a 10-year review. Transfusion 2008;48: 1699-706. 9. Bakchoul T, Meyer O, Agaylan A, et al. Rapid detection of HPA-1 alloantibodies by platelet antigens immobilized onto microbeads. Transfusion 2007;47:1363-8. 10. Chong W, Metcalfe P, Mushens R, et al. Detection of
CONFLICT OF INTEREST
human platelet antigen-1a alloantibodies in cases of
VR and MM are employees of Immucor-GTI Diagnostics and
fetomaternal alloimmune thrombocytopenia using recombinant β3 integrin fragments coupled to fluorescently labeled beads. Transfusion 2011;51:1261-70.
HOLOGIC, Waukesha, WI. The other authors have no conflict of interest.
REFERENCES 1. Serrarens-Janssen VM, Semmekrot BA, Novotny VM, et al. Fetal/neonatal allo-immune thrombocytopenia (FNAIT): past, present, and future. Obstet Gynecol Surv 2008;63:23952. 2. Hod E, Schwartz J. Platelet transfusion refractoriness. Br J Haematol 2008;142:348-60. 3. Shtalrid M, Shvidel L, Vorst E, et al. Post-transfusion purpura: a challenging diagnosis. Isr Med Assoc J 2006;8: 672-4. 4. von dem Borne AE, Verheugt FW, Oosterhof F, et al. A simple immunofluorescence test for the detection of platelet antibodies. Br J Haematol 1978;39:195-207. 5. Kiefel V, Santoso S, Weisheit M, et al. Monoclonal antibody-specific immobilization of platelet antigens
11. Skaik Y, Battermann A, Hiller O, et al. Development of a single-antigen magnetic bead assay (SAMBA) for the sensitive detection of HPA-1a alloantibodies using tagengineered recombinant soluble β3 integrin. J Immunol Methods 2013;391(1-2):72-80. 12. Fujiwara K, Shimano K, Tanaka H, et al. Application of bead array technology to simultaneous detection of human leucocyte antigen and human platelet antigen antibodies. Vox Sang 2009;96:244-51. 13. Colombo MB, Haworth SE, Poli F, et al. Luminex technology for anti-HLA antibody screening: evaluation of performance and of impact on laboratory routine. Cytometry B Clin Cytom 2007;72:465-71. 14. Décary F, Vermeulen A, Engelfriet CP. A look at HLA antisera in the indirect immunofluorescence technique (IIFT). In: Kissmeyer-Nielsen F, editor. Histocompatibility testing. Copenhagen: Munksgaard; 1975. p. 380-90.
Volume **, ** **
TRANSFUSION
7
