Erythrocyte-magnetized technology (EMT;

Transcription

1 IMMUNOHEMATOLOGY Erythrocyte-magnetized technology: an original and innovative method for blood group serology Olivier Bouix, Virginie Ferrera, Maryvonne Delamaire, Jean Claude Redersdorff, and Francis Roubinet BACKGROUND: Erythrocyte-magnetized technology () is a new fully automated method for ABO-RH-K phenotyping and antibody detection. The magnetization of red cells avoids centrifugation and washing phases. This report describes the results of an evaluation of this new technology on its specific automated system. STUDY DESIGN AND METHODS: ABO-RH-K phenotyping was compared between and a semiautomated routine method (liquid microplate for ABO-D and microcolumn system for RH-K) on 311 patients samples. The overall performance of the new method was further assessed in daily routine on a total of 11,022 samples during 3 months in two different laboratories. Antibody detection was evaluated on 624 consecutive patients samples and on 118 frozen samples containing specific antibodies in comparison with commercial microcolumn systems. RESULTS: Eight of 311 ABO-RH-K tests (2.6%) were not interpreted by. Seven of them were weak antigen or reverse grouping reactions showing a negative result with the routine method. On a 3-month follow-up, 216 of 11,022 tests (1.96%) were not interpreted by the system, 75 percent of them being due to weak or mixed-field reactions. was better in detecting ABO-D mixed-field reaction than routine microplate method. Detection of clinically significant antibodies was similar between and microcolumn. In contrast, detected a markedly lower rate of presumed nonsignificant antibodies. The system presents an overall high reliability. CONCLUSION: is tailored to meet the needs of the transfusion service and represents an important advance in the field of immunohematology. Erythrocyte-magnetized technology (; Diagast, Loos, France) is an original and innovative technology based on the magnetization of red blood cells (RBCs). The use of magnetic beads to allow quick transfer of antigens or antibodies with simple magnets has been an innovation force for automated diagnostic devices in fields other than immunohematology. This principle allows a highly reliable automation without the need for centrifugation. 1 The heart of is the adsorption of paramagnetic (magnetism occurring only in the presence of an externally applied magnetic field) compounds on the membrane of the RBCs. Thus, when placed on a magnetic plate after contact with antibodies, reactive and nonreactive magnetized RBCs are rapidly pulled to the bottom of the well. A final phase of shaking reveals or negative reactions. is running on the fully automated workstation (QWALYS 2, Diagast), a Tecan platform specially designed for this technology. The system ensures automatic handling of the whole analytic process, from the primary tube up to the final result, including bidirectional communication with the laboratory information system. In this work, we report the technical characteristics and performances of the QWALYS 2 and the results of an evaluation of for ABO-RH-K phenotyping and antibody detection. ABBREVIATIONS: = erythrocyte-magnetized technology; IQC(s) = internal quality control(s). From the French Blood Establishment, Nimes, Marseille, Rennes, and Tours, France. Address reprint requests to: Olivier Bouix, MD, PhD, Etablissement Français du Sang, CHU Caremeau, CS 68223, Nimes, France; olivier.bouix@efs.sante.fr. Received for publication January 27, 2008; revision received March 12, 2008, and accepted March 18, doi: /j x TRANSFUSION 2008;48: TRANSFUSION Volume 48, September 2008

2 ERYTHROCYTE-MAGNETIZED TECHNOLOGY MATERIALS AND METHODS ABO-RH-K phenotyping Fig. 1. Photograph of a premagnetized RBC used as reagent for reverse ABO grouping or antibody detection. Magnetic polymer beads are adsorbed on the cell membrane (arrows). We used a kit (DuoLys, Diagast) for ABO-RH-K phenotyping on QWALYS 2. The kit contains ready-to-use microplates with predried monoclonal antibodies (MoAbs) in a storage medium. The following immunoglobulin M (IgM) antibody clones were used: anti-a, 9113D10; anti-b, 9621A8; anti-a,b, 9113D D12; anti-d, P3X61; anti-c, P3X25513G8 + MS24; anti-e, 906; anti-c, MS33; anti-e, P3GD512 + MS63; and anti-k, MS56. Each microplate enables the determination of eight forward and reverse ABO grouping and eight RH-K phenotyping tests. The patients RBCs (6 ml) are first mixed with 144 ml of magnetizing solution of iron chloride (MagneLys, Diagast) and 450 ml of bromelin (Diagast) in a dilution plate. Then 40 ml of the cell suspension is dispensed in microplate wells containing predried antisera. For reverse ABO grouping, 25 ml of premagnetized 1 percent suspension A1 and B group reagent RBCs (HemaLys, Diagast) are dispensed in two empty wells and mixed with 25 ml of patient s plasma. After gentle shaking and incubation phase of 10 minutes at room temperature, the microplate is placed on a magnetic plate, and magnetized RBCs rapidly migrate and form a pellet at the bottom of each well. After being shaken, the free RBCs are resuspended while agglutinated cells form a button in the center of the well. A reader with CCD sensor integrated into the workstation displays an on screen picture of the selected microplate for validation of results by the operator. Semiquantitative results are graded from negative to 4+ depending on the threshold values defined in the software. Undetermined results are flagged as a question mark and may correspond to weak reactions, dual population of RBCs, presence of bubbles, or absence of reagent or test RBCs in the well. Results of ABO-RH-K phenotyping with QWALYS 2 on 311 consecutive patients blood samples were compared to those obtained with a semiautomated workstation (ABS Precis 3001, Immucor, Inc., Norcross, GA) using liquid MoAbs reagents on microplate (Immucor) and monoclonal RH/K reagents (, Bio-Rad, Marnes la Coquette, France) for ABO-D and RH-K phenotyping respectively. The following IgM antibody clones were used: anti-a, BIRMA-1; anti-b, F84-3D6 + F97-2D6; anti- A,B, BIRMA-1 + ES4 + ES15; anti-d, RUM-1 (Immucor); anti-c, MS24; anti-e, MS260 + MS258; anti-c, MS33; anti-e, MS63 + MS21 + MS16; and anti-k, MS56 (Bio-Rad). We systematically checked the automated interpretation of results by visual observation on screen microplate photograph. Discordant or undetermined ABO-D results were tested again by using monoclonal ABO-RH1 microcolumn reagents (, Bio-Rad) or tube test methods. Additionally, undetermined results were registered during a 3-month period of routine use in two different laboratories representing a total of 11,022 ABO-RH-K phenotyping. They were further explored using systems or manual tube tests. The capability of the method to detect dual populations of cells was studied by mixing washed RBCs from 0 to 100 percent and negative cells for each antigen studied in duplicate. Cells of appropriate phenotypes were obtained from ABO-RH-K internal quality control (IQC) reagents (French Blood Establishment, Marseille, France). We defined the percentage range of antigen- and -negative RBCs automatically interpreted as dual population, and we compared it to blinded visual interpretation. Only ABO-D type was tested for detection of dual population with the routine method. The stability of A1 and B premagnetized reagent RBCs was verified 3 days after their 3-week shelf life in triplicate using group O and AB IQC plasma samples (French Blood Establishment). Antibody detection An kit (ScreenLys, Diagast) is based on the indirect antiglobulin test (IAT) principle. The kit contains microplates coated with anti-igg antibody, a high-density solution (NanoLys) that prevents contact between patient s plasma and coated anti-igg, and a diluent (Screen- Diluent). Three vials of premagnetized group O test RBCs (HemaScreen I, II, III) are supplied separately. Magnetization is made by the manufacturer with paramagnetic polymer beads adsorbed on membranes (Fig. 1). Their phenotypes are selected in accordance with the French laboratory good practices. 2 The machine first dispenses 60 ml of NanoLys into the coated wells and then 60 ml of diluent, 15 ml of patient s plasma, and 15 ml of test RBCs (1% suspension; Fig. 2). The ScreenLys plate is incubated at 37 C for 20 minutes and then placed on the magnetic shaker. Sensitized cells migrate through NanoLys and Volume 48, September 2008 TRANSFUSION 1879

3 BOUIX ET AL. Magnetized reagent RBCs Plasma sample Screen Diluent dispensed first the and then the negative sample. The test was performed on a whole microplate with 2 16 aliquots. The stability of reagent group O cells was verified 3 days after their 3-week shelf life in triplicate using anti-d and anti-fy a IQC samples (French Blood Establishment). NanoLys Anti-human IgG Fig. 2. Principle of antibody detection. react with coated anti-igg at the bottom of the wells. Positive reaction appears as a cellular layer while negative reaction corresponds with a dot at the bottom of the well. In contrast with other solid-phase RBC adherence tests, removes the need for washing since unbound antibodies stay in the upper layer of the well and do not interact with coated anti-igg. A total of 32 samples can be analysed on each 96-well microplate. The results reported by the instrument are graded from negative to 4+ depending on the intensity of the reaction. Automated antibody detection by was compared to anti-igg/c3d and ScanCell I, II, and III reagents (Bio-Rad) on ABS Precis 3001 on 624 consecutive patients blood samples. All discrepant results were further examined manually using DiaMed-ID low-ionicstrength saline (LISS)-direct antiglobulin test and ID-DiaCell I, II, and III (DiaMed, Cressier sur Morat, Switzerland) reagents. Whenever possible, the immunoglobulin class of antibodies was determined by dilutions of plasma with dithiothreitol (DTT) solution (Sigma, St Louis, MO) or phosphate-buffered saline (Sigma) for discrepant -negative samples. Antibody identification by IAT and enzyme techniques was done manually using anti-igg/c3d, Neutral, ScanPanel, and ScanPanel P reagents (Bio-Rad). We also studied a panel of 118 frozen samples containing specific RBC antibodies previously identified using reagents. Frozen samples were thawed at 37 C and then briefly centrifuged before parallel tests with,, and DiaMed-ID reagents. The sensitivity of the method was assessed using doubling dilutions of a strong anti-fy a, two IQCs having a titer of 4 by LISS tube IAT (a weak anti-d and a weak anti-fy a ; French Blood Establishment), and a standard anti-d at 20 ng per ml (National Institute of Blood Transfusion, Paris, France). The capability of to detect IgM class antibodies was tested using commercial monoclonal IgM anti-e (clone 906) and anti-jk a (clone P3HT7) reagents. We tested the machine for carryover by using aliquots of a strong anti-d (5 mg/ml) and an inert sample placed on the sample racks so that each needle aspirated and Technical aspects of automation We assessed the hardware and software reliability of the automated system during a 3-month period of daily routine use in two of our laboratories. We recorded any downtime period of more than 30 minutes and any pipetting errors not flagged by the system. We also paid attention to variables such as management of the workload, throughput times, operator s time working on the system, and ease of routine maintenance and start-up procedures. ABO-RH-K phenotyping RESULTS Comparison with routine method Among 311 samples tested, we observed 8 (2.6%) discordant results between and our routine semiautomated method. Two reactions with anti-d reagent, one with anti-a, and four in reverse ABO grouping were flagged as not interpreted by but gave a negative result with the routine method. Those seven reactions were found weakly by visual observation of microplate and confirmed weakly reactive by manual monoclonal ABO-RH1. One transfusion related dual population with anti-e was correctly interpreted with the routine system but yielded a flagged undetermined result with. Rate and causes of undetermined results Among 11,022 samples tested in two of our laboratories during a period of 3 months, 216 (1.96%) were not automatically interpreted by the system. Extrinsic causes of undetermined results such as weak reverse grouping reaction, transfusion-related dual population of RBCs, or weak antigen expression account for 75 percent of undetermined results. Intrinsic causes were due to small bubbles or contaminating particles in a well or pipetting failure (Table 1). Detection of dual populations of RBCs (Table 2) Both automatic and visual detection of dual population of cells varied depending on the antibody reagent. The threshold percentage of antigen- cells giving a negative result was less than 20 to 30 percent for ABO and RH reagents. Positive results were observed with at least 40 to 60 percent of cells depending on the reagent. For anti-k, only the 60 percent cell mixture was detected as a dual population. was better in detect TRANSFUSION Volume 48, September 2008

4 ERYTHROCYTE-MAGNETIZED TECHNOLOGY TABLE 1. Causes of 216 undetermined results in a series of 11,022 ABO-RH-K phenotyping Extrinsic causes Intrinsic causes Cause ABO reverse Weak antigen Mixed field Pipetting error Bubbles Particles Unknown Number (%) 108 (1) 30 (0.3) 18 (0.2) 6 (<0.1) 30 (0.3) 12 (0.1) 12 (0.1) TABLE 2. Thresholds of dual population detection by, the routine method, and visual observation* Percentage of antigen- cells Reagent anti- Method A ???? - - Routine ??? Visual + +??????? - - B ???? - - Routine ??? Visual + +??????? - - AB ???? - - Routine ??? Visual + +??????? - - D ????? - - Routine ??? Visual + + +?????? - - C ???? Visual + +??????? - - E ????? - - Visual + + +?????? - - c ???? - - Visual ????? - - e ???? Visual + + +?????? - - K ? Visual + + +????? * The phenotypes of mixed IQC RBCs cells were as follows: 1) A1 D- C- E- c+ e+ K- and O D+ C+ E- c- e+ K+ for tests of anti-a, -AB, -D, -C, -c, and -K reagents and 2) A1 D- C- E- c+ e+ K- andbd+ C- E+ c+ e- K- for anti-b, -E, and -e reagents.? = detection of dual populations of cells. Thresholds of detection have been determined only for ABO-D type with routine method. Results from visual observation are expressed as (+),?, or negative (-). negative TABLE 3. Number and percentage of concordant and discordant antibody detection results between and technologies negative negative negative 532 (85.3%) 9 (1.4%) 22 (3.5%) 61 (9.8%) ing ABO-D mixed-field reactions than the routine method, but blinded visual detection always showed more accurate results. Visual checking of automated interpreted results We did not record any discrepancy between ABO-RH-K phenotypes correctly interpreted by the system and visual observation by the operator on a series of 311 samples. Weak reactions or dual population of cells not interpreted automatically, however, were usually correctly interpreted by visual checking. Stability of reagents The expected results for the two group O and AB plasma IQCs were obtained with HemaLys A1 and B cells tested 3 days after their 3-week shelf life. Antibody detection Comparison with routine method Of 624 samples tested, 83 (13.3%) and 70 (11.2%) gave a reaction in and, respectively. Details of the number and percentage of concordant and discordant results between and technologies are shown in Table 3. Among all the tests, a total of 69 antibodies were identified using technology, including mixed antibodies: anti-d (36) (28 of them were prophylactic injection in women), -C (3), -E (6), -c (3), -C w (2), -K (3), -Kp a (1), -Fy a (3), -Jk a (2), -M (3), -Le a (1), and -Le b (1).Warm autoantibodies were also present in five samples. Discrepant negative results (n = 9, 1.4%) concerned anti-d (1), -E (1), -c (1), -C w (2), -Kp a (1), -M (2), and Volume 48, September 2008 TRANSFUSION 1881

5 BOUIX ET AL. TABLE 4. Results of antibody detection on frozen plasma samples containing specific antibodies Specificity (number) DiaMed-ID Anti-D (22) Anti-C (7) Anti-E (5) Anti-c (11) Anti-K (24) Anti-Jk a (13) Anti-Jk b (3) Anti-Fy a (10) Anti-Fy b (2) Anti-M* (11) Anti-Le a * (5) Anti-Le b * (5) * Antibody specificities known as moderately or nonclinically significant. TABLE 5. Comparative sensitivity between, microcolumn systems, and LISS tube IAT using doubling dilutions of specific alloantibodies Alloantibody titer titer DiaMed-ID titer LISS tube IAT titer Anti-D Anti-K Anti-Fy a Anti-Fy a Sensitivity of The results of tests for sensitivity of are shown in Table 5. Sensitivity assays using a standard 20 ng per ml anti-d yielded a detection threshold between 1.25 and 2.5 ng per ml with both and reagents. -Le a (1). The Kp a antigen was not present on the test RBC panel. The anti-d, -E, -C w and -M not detected by were confirmed to be IgM class antibodies by DTT inactivation. The two others (1 anti-c and 1 anti-le a ) showed very weak reactions with and were not detectable with DiaMed-ID reagents. DTT inactivation yielded inconclusive results due to the weakness of the two antibodies. Discrepant - results were observed in 22 (3.5%) cases. Twelve of them were observed with plasma samples with hemolysis, icterus, or excess of lipids. The remaining 10 showed weak but apparently specific reactions which were not confirmed by or DiaMed-ID reagents or by IAT identification reagents. Two of them, however, were weakly a few days before with system (prophylactic anti-d) and were reactive with the two D+ cells. Two others gave weak unidentified reactions in enzyme techniques for identification. Analysis of frozen plasma samples with specific antibodies and commercial IgM antibodies Of the 118 samples tested, 97 (82.2%) were reactive with compared with 113 (95.7%) and 104 (88.1%) for and DiaMed-ID, respectively (Table 4). Results for Rh, Kell, Duffy, and Kidd blood group systems antibodies were similar between the three antibody detection technologies, each of them missing a few weak antibodies. As expected, was poor at detecting anti-m, -Le a, and -Le b. We observed a result at 1-in-16 dilution with a very strong monoclonal IgM commercial anti-e but its reactivity was much stronger with and DiaMed-ID reagents (1 in 2048 and 1 in 1024, respectively). The IgM commercial anti-jk a reagent, while reactive at titers of 256 and 128 with and DiaMed-ID reagents, respectively, was not detected at all by. Stability of reagents The expected results were obtained for anti-d and anti-fy a IQC with HemaScreen I, II, and III cells tested 3 days after their 3-week shelf life. Carryover We did not record any problem of carryover by testing successively a strong anti-d and an inert sample. Technical aspects of the automated system We did not record any episode of downtime requiring an engineer s visit during a 3-month period of use representing a total of 11,022 ABO-RH-K in two different laboratories. A problem with many small bubbles in A 1 and B cells wells occurred on 2 consecutive days in one of our two laboratories. This problem was resolved in-house by replacing the needles after advice from the company. On two occasions, we experienced problems of unknown cause with aspiration of test RBCs which were not flagged by the system. In all cases, those problems yielded undetermined results. Flagged pipetting errors were mainly due to low volume of reagents. A few clots were also detected. A few failures in samples, reagents, or microplates bar code reading were observed but all of them were due to errors of operators in positioning the objects and were flagged by the system. Since the QWALYS 2 can be used concomitantly for unique large as well as small continuously loaded batches, it is difficult to accurately assess representative throughput times with regard to daily routine use. We observed, however, an hourly throughput of 42 ABO-RH-K phenotyping or 55 antibody detection tests. First results were available in 30 and 45 minutes, respectively. In the list of test mode, approximately 30 samples randomly placed on the racks (10 ABO-RH-K, 10 antibody detection, and 10 ABO-RH-K plus antibody detection) were treated in 1 hour 10 minutes TRANSFUSION Volume 48, September 2008

6 ERYTHROCYTE-MAGNETIZED TECHNOLOGY In agreement with French laboratory good practices, 2 the system checks the results of IQCs at least once per day. Records of each individual test ensure total traceability regarding sample number, date of test, reagent lot numbers, operator, detailed results by well, and eventual manual modification of results. We estimated that operator s working time was reduced by 60 percent compared with our semiautomated routine method. DISCUSSION Immunohematology has been the field of important evolutions of the analytical methods for the past few years. 3 The almost simultaneous introduction of microcolumn and microplate agglutination technologies, together with their progressive automation and computerization, represent important advances. 4-7 Both technologies demonstrate an adequate level of sensitivity when compared with the LISS tube IAT. The main improvements brought by automation are the reduction in the risk of human errors, the guarantee of a reliable traceability of all elements involved in the analytical process, and the management of all the alarms of dysfunction of the system. In other respects, automation is also a well-adapted response to the constant increase of workload in the majority of blood bank laboratories. Today, a number of systems allow automated and computerized realization of ABO-RH-K phenotyping and antibody detection. With regard to operator s time and cost-effectiveness of the systems, however, it is important to distinguish fully automated devices that are able to handle all steps from sample loading to final results from semiautomated ones that require intervention of the operator for incubation, centrifugation, shaking, and reading phases. On the other hand, it is obvious that the more the system is equipped with complex automated parts such as centrifuge or washing station, the more it is subjected to mechanical dysfunction and subsequent downtime. Another important concern with the new automated technologies used for antibody detection is that their aim at improving performance should not be limited to the search for an overall increase in specificitysensitivity but must take into account their capability to detect the clinically significant and to avoid as much as possible the unwanted nonsignificant antibodies We evaluated for the first time the original and innovative for ABO-RH-K phenotyping and antibody detection with the fully automated microplate workstation QWALYS 2 in two different laboratories of the French Blood Establishment. The QWALYS 2 offers excellent security, comparable to other solid-phase microplate systems. Like other fully automated walkaway systems, however, it is limited to one single technology and a single manufacturer for both ABO-RH-K phenotyping and antibody detection. We noted a high level of mechanical reliability and global robustness of the system because we did not record any serious failure or downtime. itself is directly involved in the remarkable reliability of the system since the absence of centrifugation requirements and multiple washing steps needed for solid-phase antibody detection are still limiting factors in automation development. The principle of magnetization and magnetic fields that replace centrifugation is so simple that it is virtually life-guaranteed. Similarly, the use of a highdensity solution that avoids contact of RBC IgG antibodies with coated anti-igg and replaces the washing steps is also a key element in the reliability of the process. Another important variable that participates to the robustness of the workstation is the relatively limited number of pipetting steps due to the predried microplate reagents for ABO-RH-K phenotyping compared to systems using liquid reagents. Also, the stability of both reverse ABO and detection premagnetized reagent cells was successfully verified meaning that the process of RBC magnetization by polymer beads is quite reliable. Comparison of ABO-RH-K phenotyping between and our routine semiautomated method, despite a relatively small number of samples, showed higher performance of in terms of detection of weakly expressed antigens or weak reactions in reverse grouping, which were not interpreted by the system but found weakly by visual observation and confirmed weak by the more sensitive reagents. The reliability of ABO- RH-K phenotyping by was further demonstrated by a low percentage (1.96%) of undetermined results on a 3-month period of routine use. In comparison, the routine method had a rate of 4.0 percent undetermined results. The majority (75%) of -undetermined results were due to extrinsic causes like weak reverse ABO or weak antigen reactivity. Undetermined results due to the system itself or of unknown cause represented only 0.5 percent of 11,022 samples. Our laboratories are working for large university hospitals where multitransfused trauma patients often come in the emergency ward without a previously established blood typing. Thus, we were interested in evaluating for the automatic detection of experimental dual population of RBCs. The thresholds of detection of dual populations of cells by were rather satisfactory even though visual observation of microplates on screen was always superior. The detection of dual population of cells resulted in a simple flag of undetermined result with, while it was specifically marked as DP by the system. RBC antibodies are major factors contributing to the risk of RBCs transfusion and are also implicated in fetalmaternal incompatibility and hemolytic disease of the fetus and newborn. There are conflicting data in the literature about the respective sensitivity of new methods such as microcolumn and solid-phase systems compared to the gold standard tube LISS-IAT. 5,9,12 High-sensitive tests Volume 48, September 2008 TRANSFUSION 1883

7 BOUIX ET AL. systems usually produced higher rates of unwanted false and nonclinically relevant antibodies. 9 The use of the most sensitive techniques to detect weakly reactive antibodies, however, such as Kidd antibodies, which are commonly implicated in delayed hemolytic transfusion reactions, is highly recommended. 13 Our study of antibody detection by is in favor with a good performance of this method in detecting clinically relevant antibodies. We observed similar performance of and in detecting clinically relevant antibodies to Rhesus, Kell, Kidd, and Duffy blood group antigens, each detecting 92 of the 97 antibodies (95%), while DiaMed-ID detected only 87 of the 97 antibodies (90%). This is in agreement with the results of a detection threshold of 1.25 to 2.5 ng per ml anti-d for both and. On the other hand, when compared to the sensitive method, discrepant negative results observed with concerned either IgM class antibodies or very weak reactions which were not detected with DiaMed-ID reagents. detected a markedly lower number of nonrelevant or moderately relevant antibodies (anti-m, -Le a, -Le b ) routinely detected by microcolumn methods, especially in pregnant women. Although these antibodies are known to cause delayed hemolytic transfusion reactions or hemolytic disease of the fetus and newborn in rare cases, they are typically naturally occurring unwanted IgM antibodies. 14,15 Only in a few cases of anti-m, an IgG component can be detected and may cause fetal anemia or reduced survival of transfused M+ RBCs. 16,17 Those data, along with the observation that does not detect a monoclonal IgM anti-jk a and detects an IgM anti-e only at extremely high concentration compared to microcolumn reagents, show that is likely to detect the rare but potentially significant IgG class anti-m. The nondetection of unwanted IgM antibodies might be considered an advantage of the technique. We observed a relatively high rate (3.5%) of discrepant - results compared with. All of them were very weak or noninterpreted reactions. More than half (12/22) were due to an abnormal coloring of plasma samples with hemolysis, icterus, or excess of lipids and can be considered as unwanted false results. The technical instructions from the manufacturer, however, clearly state that erroneous results can be observed if the sample shows traces of hemolysis. The remaining 10 discrepant samples showed reactions with only one or two of the three test RBCs. It is of interest to note that two of them reacted with the two D+ cells and were with reagents a few days before (4 and 23 days, respectively) with an identified prophylactic anti-d. We cannot state objectively, however, that those samples contain antibodies since antibody identification has not yet been developed by the manufacturer. The availability of antibody identification test in the near future will be of special interest to further assess the sensitivity-specificity of. In summary, the new automated represents an important technologic advance in immunohematology: 1) owing to the technology itself, the simplicity of the automation makes it highly reliable and well adapted for medium- and large-sized facilities; 2) it allows handling greater workload with no increase in staff; and 3) the high level of security and the total traceability perfectly fulfills the constraints of good clinical laboratory practices. Compared with our routine semiautomated method, ABO-RH-K phenotyping showed very good performances. antibody detection appears to successfully balance sensitivity with specificity. Its comparison with microcolumn systems shows similar performance in detecting clinically relevant antibodies together with a noteworthy reduction in the number of antibodies without clinical importance. ACKNOWLEDGMENTS The authors thank Marie-Christine Calloix, Julia Gouvitsos, and Veronique Godefroy for collection of data, scientific advice, and critical review of study proposal. They are also grateful to Nathalie Peyric and Sylvie Flageul for technical assistance. REFERENCES 1. Elaissari A, Veyret R, Mandrand B, Chaterjee J. Biomedical application for magnetic latexes. Surfactant Sci Ser 2003; 116: Arrêté du 26 avril 2002 modifiant l arrêté du 26 novembre 1999 relatif à la bonne exécution des analyses de biologie médicale. J Officiel de la Republique Francaise 2002;26: Voak D. New developments in blood group serology. Infusionsther Transfusionsmed 1999;26: Bruce M, Hoppe PA, Kochman SA, Le Pennec PY, Moore PB, Voak D. A report: Reagents for the 1990 s. lmmunohematology 1991;7: Voak D. The status of new methods for the detection of red cell agglutination. Transfusion 1999;39: Lapierre Y, Rigal D, Adam J, Josef D, Meyer F, Greber S, Drot C. The gel test: a new way to detect red cell antigenantibody reactions. Transfusion 1990;30: Plapp FV, Sinor LT, Rachel JM, Beck ML, Coenen WM, Bayer WL. A solid phase antibody screen. Am J Clin Pathol 1984;82: Casina TS. In search of the holy grail: comparison of antibody screening methods. Immunohematology 2006;22: Weisbach V, Kohnäuser T, Zimmermann R, Ringwald J, Strasser E, Zingsem J, Eckstein R. Comparison of the performance of microtube column systems and solid-phase systems and the tube low-ionic-strength solution additive 1884 TRANSFUSION Volume 48, September 2008

8 ERYTHROCYTE-MAGNETIZED TECHNOLOGY indirect antiglobulin test in the detection of red cell alloantibodies. Transfus Med 2006;16: Garratty G. Screening for RBC antibodies what should we expect from antibody detection RBCs. Immunohematology 2002;18: Garratty G. How concerned should we be about missing antibodies to low incidence antigens? Transfusion 2003;43: Bromilow IM, Adams KE, Hope J, Eggington JA, Duguid JK. Evaluation of the ID-gel test for antibody screening and identification. Transfus Med 1991;1: The Serious Hazards of Transfusion Steering Group. SHOT Annual Report 2005;1-84. Available from shotuk.org/ 14. Judd J. How do I manage cold agglutinins. Transfusion 2006;46: De Young-Owens A, Kennedy M, Rose RL, Boyle J, O Shaugnessy R. Anti-M isoimmunization: management and outcome at the Ohio State University from Obstet Gynecol 1997;90: Wilkman A, Edner A, Gryfelt G, Jonsson B, Henter JI. Fetal haemolytic anemia and intrauterine death caused by anti-m immunization. Transfusion 2007;47: Furukawa K, Nakajima T, Kogure T, Yazaki K, Yoshida M, Fukaishi T, Ibuki Y, Igarashi M. Example of a woman with multiple intrauterine deaths due to an anti-m who delivered a live child after plasmapheresis. Exp Clin Immunogenet 1993;19: Volume 48, September 2008 TRANSFUSION 1885

9