How concerned should we be about missing antibodies to low incidence antigens?

Antibodies to low incidence antigens (ab LIA) have been shown to sometimes cause hemolytic transfusion reactions (HTRs) and hemolytic disease of the fetus/newborn (HDFN). For many years “complete” compatibility testing included an antibody screen and a crossmatch that included an antiglobulin test (AGT). The crossmatch usually detected ab LIA if a donor happened to possess a LIA. In the late 1970s, the “type and screen” approach was suggested for surgical patients who rarely required blood transfusion.1–4 This approach used the AGT for the antibody screen, but not for the crossmatch. Data were published to show that although some antibodies might not be detected, the number of clinically significant antibodies missed was small. Because the surgical patients selected for type and screen were usually not transfused, the approach was widely accepted without much controversy. The controversies started when, in the 1980s, it was suggested that this approach could be used for all patients requiring transfusion.4,5 Concerns were raised that antibodies of potential clinical significance could be missed for several reasons: 1) ab LIA would not be detected; 2) antibodies showing dosage (e.g., anti-Jka) could be missed if antibody detection RBCs were not from homozygotes [e.g., Jk(a + b–)]; 3) antibody detection RBCs may not have antigens such as f(ce) [e.g., R1R1 and R2R2 RBCs will not be f+ but crossmatched donor units (e.g., rr) may have the antigen]; and 4) weak antibodies may react with RBCs from fresh donor units but may not react with the older antibody detection RBCs. In 1984, regulatory authorities in the US allowed the crossmatch to be abbreviated (i.e., “type and screen” approach) for patients requiring transfusion. This decision was based on data showing a cost-benefit and the acceptable safety of this approach. For instance, Table 1 shows results from seven studies3,7–12 which indicate that the risk of missing an antibody of potential clinical significance, by using an abbreviated crossmatch, was 1 in 5494 recipient sera, or 1 in 10,615 crossmatches. Table 2 shows the antibodies that were detected by the crossmatch but not by the antibody screen. It is interesting to note that only about half of the 75 antibodies were directed against antigens that usually are not present on antibody detection RBCs (e.g., LIAs). Judd13 found that 220,000 antiglobulin crossmatches yielded 35 potentially clinically significant antibodies that were not detected by the antibody screen (a risk of 1/6286 crossmatches). Only six of these were ab LIAs (two anti-Jsa, two anti-V, one anti-Cw, and one anti-Wra). Thus, the risk for missing an ab LIA was 1 per 36,666 crossmatches. It is interesting to note that 17 of the missed antibodies were Rh antibodies and the other 12 were anti-K, -Jka, -Jkb, and -Fya. It is probable that these antibodies may have been detected if the antibody detection RBCs had always come from donors homozygous for the putative antigens. Although ab LIA may not be detected by the abbreviated crossmatch, the real risk to a patient is a unit that possesses a putative antigen and whether the ab LIA is clinically significant. The incompatibility frequency (antigen frequency X antibody frequency) has been calculated to range from 0.000100 for Cw to 0.000017 for Jsa (similar to Wra).6,14 In 1990, Shulman15 gathered data from 20 hospitals in different parts of the United States. These hospitals had issued blood for transfusion based on 1.3 million negative antibody screens and negative immediate spin crossmatches. Only five patients experienced an acute overt HTR (a risk of one overt HTR per 260,000 crossmatches). The antibodies associated with the HTRs, and undetected by the antibody screen, were one example each of an anti-Jka, -C, -c, -Wra, and -Kpa. Thus, the risk of an overt HTR associated with ab LIA was 1 per 650,000 crossmatches. Thus, even if an ab LIA is present, the risk of a patient receiving an incompatible unit is very small. In addition, ab LIAs are often clinically insignificant, being IgM and/or not reacting at 37°C. In the US, more than 150 million units of blood must have been transfused using abbreviated crossmatching procedures utilizing antibody detection RBCs without specially selected LIAs. There are no data I know of that indicate that this has been an unacceptable approach. The risks discussed above are pertinent to the US and probably Europe, but may be different in other countries. For instance, the most common antibody detected in Hong Kong, Taiwan, and Thailand is anti-Mia, with an incidence of 0.3 percent and 0.5 percent in patients and pregnant women, respectively.16 The MiIII phenotype that reacts with anti-Mia occurs in approximately 7 percent of Chinese as compared to 0.012 percent of Caucasians.16 With such a high incompatibility frequency, it has become common practice in Taiwan and Hong Kong to include the MiIII phenotype in antibody detection RBCs.17,18 However, because many anti-Mia appear to be clinically insignificant, it has been argued by some Chinese investigators that this approach has no cost-benefit.19 The abbreviated crossmatch has not been used as commonly in Europe (e.g., the UK) as in the US during the last two decades. For instance, according to survey results from the year 2000, 88 percent of hospitals in the UK were still using the AGT in their crossmatch (i.e., a “complete” crossmatch),20 compared with less than half of the hospitals in the US. The electronic crossmatch is associated with similar risks as the “type and screen” approach because the antiglobulin test is only performed at the antibody screen phase. Because of the increasing use of the electronic crossmatch, there seems to have been more concern about missing ab LIA during the last few years in Europe than in the US. In Europe and Asia, suppliers of antibody detection RBCs (commercial reagent companies and blood centers) have faced increasing demands for antibody detection RBCs to possess some LIAs (especially Cw, Wra, and Kpa). It is difficult and expensive to supply a set of RBCs that are from donors homozygous for C, c, E, e, Jka, Jkb, Fya, Fyb, S, s and also cover Cw, Wra, and Kpa. I can only believe that some of the people wanting such RBCs may be unfamiliar with the older data showing the safety of the abbreviated crossmatch (which also apply to the electronic crossmatch). If one takes into account the clinical significance of ab LIAs, in addition to incidence of the antibody and antigen, then it is obvious that the risk to the recipient is extremely small and cost-benefit calculations should convince most people that one does not need to insist that antibody detection RBCs possess Cw, Wra, and Kpa. In addition to the publication by Shulman15 mentioned previously, other publications are pertinent to the safety of using antibody detection RBCs lacking LIAs. In a 19-year period, the Mayo Clinic found that only 4 of 197 HTRs were associated with ab LIAs (3 anti-Cw, and 1 anti-Wra).21,22 One HTR was caused by anti-Lua. Although Lua is not officially a LIA, Lua is often not present on antibody detection RBCs. Rouger23 reported that, in a 30-year period involving 600,000 antibody screens, he has only encountered two HTRs due to ab LIAs. If we consider only anti-Wra, which is the most common ab LIA in Europe and the US, occurring in about 2 percent of random sera, HTRs are rare. De La Rubia et al.24 encountered only one mild HTR in a 5-year period when 100,000 antibody screens were performed. Arriaga et al.25 reported that in a 10-year period, when 300,000 units of RBCs were transfused, only three HTRs due to anti-Wra were encountered. In this issue of TRANSFUSION, Schonewille et al.26 add to the European experience by reviewing data from their hospital from 1978 to 2000; complete crossmatches were performed before 1991. They were particularly interested in abLIA that might appear in patients who had already made other alloantibodies. Their definition of LIAs was broader than the usual definition of less than 1 percent: they used less than 10 percent (e.g., to include antigens such as Lua that are often not present on antibody detection RBCs). They reviewed the records of 1795 patients. The sera of these patients contained 2257 antibodies, and 89 patients had 94 abLIAs. Anti-Wra was the most common ab LIA. In a prospective study, 12.3 percent of the patients were shown to have anti-Wra. Thirty-nine patients had ab LIAs together with other antibodies; 20 of these patients had autoantibodies. Fifty-one ab LIAs were found as solitary antibodies in 50 patients; 14 were detected by crossmatching when a complete crossmatch was used (before 1991); 37 were detected by antibody screening tests. Six hundred and sixty-four patients who had alloantibodies detected received 7792 transfusions. Only one anti-Wra was detected by crossmatching; anti-Cw was detected posttransfusion in an eluate from DAT-positive RBCs but the patient had no signs of a HTR. During the study period, 5 percent of patients were found to have ab LIAs but no HTRs were reported as being due to ab LIAs. The authors wisely concluded that the risk to recipients was no different in the period when the abbreviated crossmatch was used to the period where a complete crossmatch was performed. This European study agrees with data generated in the US that suggest that the risk of an obvious HTR due to ab LIAs in patients who receive blood crossmatched using an abbreviated method (i.e., no antiglobulin test and crossmatch phase), and using antibody detection RBCs not especially selected to have LIAs, is somewhere in region of 1 in 500,000. Such a risk hardly justifies the expense and trouble in supplying antibody detection RBCs containing Cw, Wra and Kpa. Blood transfusion services demanding such RBCs must realize that the choice to use an abbreviated crossmatch (including an electronic crossmatch) is based on cost-benefit and risk calculations, and people who remain uncomfortable with this should use “complete” compatibility testing.