ABO Mistyping of cis-AB Blood Group by the Automated Microplate Technique

ABO blood group is known to be most important in blood transfusions; correct ABO grouping is the most important step of pre-transfusion testing to ensure safe transfusion to the patient. Considering its importance, manual tests can be of concern as they are prone to human errors. In addition, the labor-intensive nature of manual serologic tests is often considered a burden for efficient laboratory administration. Thus, clinical laboratories have been using automated systems to manage blood banks. Automated laboratory testing has several advantages, including increased quality of pre-analytical steps and reduced error rates [1]. Despite slow progress compared to that of the Western Hemisphere, automated systems for blood banks are beginning to be introduced in Korea.

It is expected that automation can be safe because of the use of bar-coded samples, lack of performance error during the analytical phase, and absence of errors in interpreting or reporting results during post-analytical phase [2]. However, we have observed some mistyping in cis-AB phenotype samples. Cis- AB is the most frequent among ABO subgroups in Korea, and, depending on the co-inherited ABO allele, phenotypes of cis- AB varies from typical A₂B₃ (when paired with an O allele), A₂B (when paired with a B allele), and A₁B₃ (when paired with an A allele) to atypical A_intB₃, A₁B_m, A₁B_x[3]. Nine cis- AB alleles (cis-AB01, cis-AB01var, cis-AB02 to cis-AB08) have been reported so far [4], and correct ABO typing in individuals is of importance, as transfusion of AB type blood to these patients can cause hemolytic adverse transfusion reactions [5,6,7]. To prevent ABO mistyping in cis-AB patients, our hospital has applied a combination of a manual tile method with automated devices. Herein, we share our experience with cis-AB phenotype cases that automated devices failed to identify properly.

The cases of this study were seen at our hospital between April 2015 and May 2016. A total of 51,495 were tested for ABO typing. After introduction of the automated devices Galileo NEO (Immucor Gamma, Norcross, GA, USA) and QWALYS-3 (Diagast, Loos, France) at our institute, they were initially evaluated and then used routinely. The sera used for cell typing in QWALYS-3 were anti-A (9113D10 clone) and anti-B (9621A8 clone), and the Galileo NEO used anti-A (F98 7C6 clone) and anti-B (F84 3D6; F97 2D6 clone).

Currently, all tests done by automated devices are duplicated by the manual tile method in our hospital. This involves mixing undiluted EDTA blood with anti-A (MH04 and A303 clone), anti-B (NB1.19, NB10.5A5 and NB10.3B4 clone), and anti-D (MAD2 clone) BioClone reagent (Ortho Clinical Diagnostics, Raritan, NJ, USA) on an acrylic tile plate. We have followed the manufacturer‘s instructions regarding the manual slide method with some modification. We omit the initial washing procedure and apply the test on an acrylic tile plate instead of slides. Reaction is interpreted with the naked eye. Examples of visual results of this method are presented in figure 1. The manual tile method is done only on cell typing with anti-A and anti-B. Weak reaction to anti-A or anti-B was defined as weaker agglutination than normal RhD-positive-to-anti-D agglutination. Cases suggestive of a cis-AB phenotype by the manual tile method were selected.

We selected cases that were flagged by the automated device, or cases that showed discrepancy between manual and automated methods. Typical cases of discrepancy of automated and manual methods are observations suggestive of decreased reactivity to anti-B in the manual tile method compared to the automated device results or weak agglutination with B-cell RBCs in manual tube method. The outline of the work process during this period is presented in figure 2.

In cases that showed discrepant results by an automated device and the manual tile method, the manual tube method was applied to confirm ABO typing. Cases of interest were tested by both automated devices. In cases of the AB blood type, forward typing with anti-A1 (anti-A1 lectin, Ortho Clinical Diagnostics) was additionally performed to confirm A₂B or A₂B₃ phenotypes, which are typical cis-AB blood phenotypes in Korea [8,9,10]. Reagents used in serologic tests were as follows: forward typing was performed using Anti-A, Anti-B, and Anti-D BioClone reagent (Ortho Clinical Diagnostics) by both tile and tube method. Reverse typing was performed using Affirmagen A1 and B cells (Ortho Clinical Diagnostics) by the tube method.

DNA was extracted from each sample with the Qiagen DNeasy Kit (Qiagen, Hilden, Germany), and ABO exons 6 and 7, along with their flanking intron sequences, were amplified and sequenced in selected samples to confirm the cis-AB blood grouping by previously described methods [10]. Sequences were analyzed using Sequencher 5.0 (Gene Codes Corp, Ann Arbor, MI, USA) software. We referred to The Blood Group Antigen Gene Mutation Database (www.ncbi.nlm.nih.gov/gv/mhc/xslcgi.cgi?cmd = bgmut/home) for nomenclature of the detected cis-AB alleles [4].

We identified 13 cases of discrepancies between the automated and manual methods, as presented in table 1. 87.5% (7/8) of cis-AB samples tested by QWALYS-3 were interpreted as AB type without any flag as an inconclusive ABO type, and 70.0% (7/10) of cis-AB samples were typed as AB by Galileo NEO. Overall, the manual methods showed decreased reactivity to anti-B sera in forward typing compared to that of the automated devices. QWALYS-3 and Galileo NEO flagged one case and two cases, respectively, as inconclusive ABO types, and each automated device interpreted 7 cis-AB cases as type AB. These cases were revealed to be ABO*cis-AB01 haplotype with ABO*O01 or ABO*O02. The overall frequency of cis-AB during this period was 0.025%.

We have observed 1 case of cis-AB with a pairing haplotype other than O allele. This was case 13 in table 1, with genotype of ABO*cis-AB01/ ABO*A102. The QWALYS-3 device showed 4+ reactivity to anti-A sera and anti-B sera in cell typing, with no anti-A or anti-B in her serum, and the case was interpreted as an AB phenotype, while forward typing by Galileo NEO indicated 4+ reactivity to anti-A sera, but did not show reactivity to anti-B sera. Concurrent evaluation by the manual tile method showed similar results when compared with the Galileo NEO; the sample was further evaluated with manual tube testing which also showed a similar phenotype as the Galileo NEO. After transfusion with two packs of A+ leukocyte-reduced RBCs, QWALYS-3 testing showed an indeterminate result against anti-B sera, whereas results of other methods showed no change in anti-B reaction. Although the exact cause of the extra-reactivity to anti-B sera in cell typing remains unsolved, ABO mistyping as typical AB with QWALYS-3 was obvious, based on results prior to transfusion (table 2).

This study illustrates the problems when automated devices are applied in a cis-AB phenotype-prevalent area. Although a rare subtype, cis-AB is encountered more frequently in Korea [10,11] and Japan [8] than in other populations [12]. A Korean study reported a case of atypical delayed hemolytic transfusion reaction in a patient with cis-AB blood type following transfusion of Rh A-positive packed red blood cells and fresh frozen plasma [7]. Since the recommended blood for transfusion to cis-AB patients is type O red cells with type AB platelets and AB plasma, it is important to discriminate cis-AB from typical AB blood [5,6,7]. In addition to transfusion concerns, there have been several paternity disputes in Korea and Japan because of an apparent contradiction to the general Mendelian inheritance of ABO blood groups [9,10]. Correct ABO typing to the subgroup level is important in that matter, and, although sometimes challenging to novice personnel, serological testing can discriminate cis-AB [13,14,15,16,17,18,19], and other ABO subgroups [20,21] using manual methods (table 3).

Our hospital encountered problems with QWALYS-3 in detecting cis-AB subtypes when initially evaluating the performance of this device [22]. QWALYS-3 interpreted multiple cases of cis-AB as typical AB; this was the initiation of this study. During our evaluation, we found that the Galileo NEO is subject to similar errors. We attempted to solve this problem by applying anti-A1 sera in cases of AB phenotype while forward typing with the tube method, and in cases with a negative reaction to anti-A1, we also performed ABO serum typing by the standard tube method, with a prolonged incubation time of 15 min at room temperature. We sought to ensure the detection of weak B-cell reaction in reverse typing.

Galileo NEO detects hemagglutination in microplates to determine the ABO blood type. This device has shown good results in ABO typing [23,24], yet the cited studies found some discrepancies between results from this device and manual methods. Discrepancies were usually a result of weak reactions in manual testing; both studies indicated that this device is more sensitive in detecting agglutination reactions. The QWALYS system utilizes erythrocyte-magnetized technology [25], and its performance in ABO blood typing has been reviewed by many researchers. The QWALYS-2 device was examined by Schoenfeld et al. [26], who concluded that it was suitable for ABO grouping. Multiple Korean domestic studies reported that a later version of this device, QWALYS-3, showed good concordance with manual methods and determined that it could be used in routine pre-transfusion testing in the blood bank [22,27,28]. However, our institution previously reported that a cis- AB sample was ABO mistyped by QWALYS-3 [22]. Thereafter, to prevent ABO mistyping, our hospital has applied a combination of testing by the manual tile method and examination with automated devices (fig. 2).

The cis-AB cases identified in this study show that both automated devices are more sensitive in forward typing than the manual methods but regarding reverse typing, QWALYS-3 and Galileo NEO showed inconsistent results. All of our cases were suspected to be AB subgroups with a cis- AB haplotype. We observed that Galileo NEO failed sporadically to detect reactions with B cells in reverse typing. Using a simple manual method, such as the tile method that was applied in this study, these shortcomings of the automated devices can be overcome. Whether or not there are inconsistencies regarding other ABO subgroups is unclear thus far; however, with the exception of cis-AB we did not detect any other systematically problematic phenotypes.

Case 13 in our study raised another interesting point. Although Galileo NEO and manual testing showed an A1 phenotype with low anti-B titer, this patient's red cells were typed as typical AB type by QWALYS-3. The phenotype of this cis-AB sample might be A type with low titer of anti-B in her serum. The missing or very low expression of the B antigen is in contrast with other typical cis-AB subjects, and this can be due to allele competition in cases of cis-AB/O genotypes. As the phenotype of a ABO*cis-AB01/ABO*A102 subject was previously observed as A1 phenotype multiple times, this lack of B antigen expression (phenotypically A1) is to be considered a not rare phenomenon in ABO*cis-AB01/ABO*A102 subjects [29,30,31]. This observation can be explained by the fact that the A1 glycosyltransferase is in competition with that from cis- AB01 and does not leave enough H substrate to produce detectable B antigen.

The present study summarizes our experience with ABO mistyping by automated microplate devices. All cases with ABO mistyping in our study had cis-AB haplotypes; thus, we suggest that automated microplate-based devices should be used with caution in cis-AB-prevalent areas. Retesting automated device results of phenotypically AB subjects that show no reactivity to anti-A1 using the tile method is a suitable approach identify cis-AB cases and to avoid mismatched transfusions in these cases.

The authors declare no conflicts of interest relevant to the publication of this article.

Sejong Chun and Mi Ra Ryu contributed equally to this work.