|
American Red Cross Biomedical Services
Medical and Scientific Updates Selecting Platelets for Transfusion of the Alloimmunized Patient
Scott Murphy, MD Prepared and Distributed by Holland Laboratories and the Medical Office Introduction Heavily transfused patients commonly become refractory to platelet transfusions. This paper will review our experience supporting such patients in the Penn-Jersey Region of the American Red Cross where we serve 90 hospitals and a population of 5 million people. This is not a prospective, randomized study and has the defects associated with retrospective analysis. However, the review has led us to some conclusions which may be helpful. Human leukocyte antigens, HLA, are expressed on glycoproteins which are integral membrane proteins present in many cells. Almost all cells have Class I antigens (A, B and C subloci) whereas only selected cells (dendritic cells, monocytes, and subsets of B cells) have Class II antigens. Primary alloimmunization to Class I HLA antigens appears to require presentation of such antigens on cells that also express Class II antigens1. Despite ongoing basic research, we still do not understand the precise details of antigen presentation. There appear to be co-stimulatory molecules on cells which bear Class II antigens. It is unclear whether it is the Class II antigens themselves or the co-stimulatory molecules which are crucial for antigen presentation2. However, at least in principle, since red cells and platelets lack Class II antigens, they should not be able to stimulate primary alloimmunization on their own. Therefore, contaminating leukocytes are implicated. For this and other reasons, transfusion of leukoreduced blood components has become quite popular. The recently completed TRAP (Trial to Reduce Alloimmunization to Platelets) trial confirmed the value of such leukoreduction in reducing the frequency of alloimmunization3. In spite of the widespread use of leukoreduction in our region (Penn-Jersey), requests for matched platelet products have remained constant over the last 10 years, approximately 200-250 products per month. There are several potential explanations for this. Patients may receive transfusions which have not been leukoreduced at small hospitals before they are referred to major centers. Leukoreduction may be much less effective if the patient is already sensitized4. Furthermore, most of the leukoreduction in our region is accomplished by bedside filtration. This may be an inadequate method for consistently removing leukocytes. Over the last 2-3 years, many of our large hospitals have moved to blood products that have been leukoreduced prior to storage. It will be interesting to see whether this will correlate with a decrease in requests for matched products. Clinical Refractoriness to Platelet Transfusion It is important to emphasize that clinical refractoriness to platelet transfusion is not necessarily due to alloimmunization. Bishop et al.5 performed multiple linear regression analyses of a variety of clinical and laboratory factors to look for correlations with corrected count increments (CCIs) to platelet transfusion. The presence of HLA antibody was definitely a powerful predictor but not as powerful as enlargement of the spleen, concomitant use of amphotericin, disseminated intravascular coagulation, or a recent allogeneic bone marrow transplantation. In the last 7 years, there have been five more publications6-10 describing correlations between clinical and laboratory factors and clinical response. HLA alloimmunization is important in all of these reports but each study suggested the importance of other non-immune factors. In four of the five, fever and bacterial sepsis were associated with a poor response and, in two of the five, there was no correlation between concomitant amphotericin use and clinical response. Thus, it appears that the clinical setting and patient selection may have a major impact on clinical response in any individual study. An important corollary point is that a matching service is usually not helpful if there is no antibody present. It has been our experience that many clinicians have inappropriate negative impressions about a matching program because "HLA matched" products are dispensed without an appropriate serologic study to demonstrate the presence of antibodies. Methods Used to Support the Alloimmunized Patient Methods used to support alloimmunized patients in the United States and throughout the world vary widely depending on the region. For example, in a recent survey of eight regions of the American Red Cross, matching according to HLA criteria and crossmatching were widely used. However, in one region more than 99 percent of the products were dispensed according to HLA criteria with only an occasional use of crossmatching. In another region, 94 percent of products were distributed solely on the basis of crossmatching with only 6 percent distributed according to HLA criteria. The other six regions displayed intermediate levels of dependence on these two methods. The Penn-Jersey Region has a very reliable HLA laboratory which can not only obtain HLA types on donors and patients but also identify HLA antibody and the antigens to which the antibody is directed. The region also has an established pool of 3,500 HLA typed donors and access to 30,000 more donors at Red Cross Regions on the east coast of the United States. Therefore, we find it convenient to use methods of support which are HLA based, supplemented with crossmatching. Furthermore, we find such methods to be effective. Antiglobulin-augmented lymphocytotoxicity (LCT) assay. In the LCT assay, 58 wells containing lymphocytes from donors with an appropriate heterogeneity of HLA types are incubated with the patient's serum and complement. Lymphocytotoxicity is assessed microscopically. With this method, one can determine whether a patient has antibody, and the extent of alloimmunization can be expressed as the percentage of cells in the tray that lyse. The percent of reactive antibody (PRA) can be from 2 percent to 100 percent. Finally, by analyzing which cells lyse, one can identify the antigenic specificity of the antibodies. Total reliance on the LCT assay leaves one open to the risk that platelet-specific (non HLA) antibody will be missed. The extent of this risk will be discussed subsequently. It can be a source of considerable confusion among physicians that, when we are studying a patient who is refractory to platelet transfusion, a test in which the lymphocyte is the target cell is suggested. However, the LCT assay has been the gold standard which has been relied upon for several decades. Approximately 50 percent of patients transfused with blood products which have not been leukoreduced develop antibody detected by this method after 1-8 weeks of transfusion experience11. The incidence appears to be higher (80%) in aplastic anemia than it is in acute leukemia (40%). It is a fundamental puzzle of immunology why half of the human population never become immunized no matter how many leukocytes are infused. Hogge et al12 showed that patients with no LCT antibody had good responses to randomly selected platelet transfusions whereas patients with such antibodies had very poor response such that when the PRA was greater than 75 percent, there was essentially no response to randomly selected platelets. We have recently reviewed our own local experience with the LCT assay using 136 sera referred over a 6-month period from patients felt to be refractory to platelet transfusion. First of all, 46 percent of the samples had no antibody demonstrable by the LCT technique. This reflects the fact which we have discussed that there are other reasons for failure to respond to platelet transfusion besides alloimmunization. In the 54 percent of antibody positive samples, the PRA ranged from 2 percent to 100 percent with all intervening levels of alloimmunization so that the patients could be classified as being mildly immunized (PRA 1-40%) moderately immunized (PRA 40-70%), and severely immunized (PRA > 70%). It is important to emphasize that this classification is based on breadth of alloimmunization, not strength (titer). Furthermore, many of the patients, at all levels of alloimmunization, demonstrated intra-cross-reactive group (CREG) antibodies. Using serologic methods, CREGs have been identified for the HLA antigens over the past 20-30 years13. It has been observed that, when an immunizing cell has only one antigen discrepancy with the recipient, the recipient’s serum may still display antibody against many HLA specificities. Thus, if the immunizing cell is A1, 2, B7, 8 and the recipient is A1, 2, B5, 8, the sera may display not only anti B7 but also anti B22, 27, and 40. This finding results from the fact that HLA antigens contain public epitopes shared with other antigens as well as the private epitope which defines the specificity. HLA antigens have been placed into CREGs based on these shared epitopes. These CREGs have been the basis for clinical HLA matching using antigens within the patient’s own CREGs14. However, as mentioned above, some patients have antibody to antigens within their own CREGs (intra-CREG antibody)15. In our experience and that of others16, the level of alloimmunization tends to remain constant over time in the majority of patients even as they continue to be transfused. Thus, patients who are mildly immunized tend to stay mildly immunized, patients who are moderately immunized tend to stay moderately immunized, and those who are severely immunized tend to stay severely immunized. An occasional patient, observed very early during alloimmunization, may have a major increase in the PRA over several weeks. On the other hand, approximately 25 percent of immunized patients may lose their antibody over time in spite of continuing transfusion,16 another phenomenon which we do not understand. In practice, this means that one can perform an LCT assay approximately once monthly to monitor the patient’s status without concern for major changes during that period. The solid phase red cell adherence assay (SPRCA). There is also a widely used method which employs the platelet as the target cell, the solid phase red cell adherence assay (SPRCA) which has been made available commercially as Capture-P by Immucor Inc., Norcross, GA. It is the major method used for platelet crossmatching in the United States. It detects HLA antibodies as well as antibodies to platelet specific antigens such as HPA1a (PlA1)and IgG antibodies to the A and B antigens which are present not only on red cells but on platelets as well. Briefly, the platelets to be used as targets are layered in microtiter wells and the patient’s serum is added. The wells are washed and anti-IgG-coated indicator red cells are added. If there is antibody to the platelets, the red cells form a thin film in the well whereas, if there is no antibody, they will "puddle" in a button at the bottom of the well. The major advantage of this method is that it is rapid so that a crossmatch-compatible product can be identified in 1-2 hours. It has the disadvantage of being subjective and dependent on the skill and experience of the technologist. In addition to its use for crossmatching, the SPRCA can be used for antibody screening. Platelets from 10-20 individuals selected for heterogeneity of HLA type can be used to study a patient’s serum. The results can be expressed as a percent of reactive antibody (PRA) for SPRCA just as one does with the LCT assay. We tested 298 sera from patients thought to be refractory to platelet transfusion by SPRCA and compared the results with the PRA obtained with the LCT assay. Fifty-three percent were positive in both assays and 28 percent were negative in both assays, so that there was concordance in approximately 80 percent of these cases. This suggests that most of the antibodies detected by the SPRCA are indeed HLA antibodies. It was extremely rare (less than 1%) for the SPRCA to be negative when the LCT was positive. Thus the SPRCA is a good method for screening for the presence of HLA antibody. On the other hand, 20 percent of the cases were SPRCA-positive and LCT-negative. The panels had not been selected according to ABO type. When we examined the 55 SPRCA-positive/LCT-negative cases, 47 (86%) were group O patients, and, of these 47 cases, 35 (75%) reacted only with platelets from group A or B donors, strongly suggesting that the antibodies detected were ABO antibodies. Thus, 64 percent of the SPRCA-positive/LCT-negative cases were probably due to ABO antibodies. We examined the CCIs (per mm3 per M2 per 1011 platelets) in these patients with ABO antibodies to platelet transfusions from group O donors as opposed to group A or AB donors. The mean CCI was 7,190 for group O platelets and 4,598 for platelets bearing the A or AB antigen (p = 0.067). Similarly, for group O patients who did not have ABO antibodies demonstrated by SPRCA, the mean CCI was 7598 in response to platelets from group A or AB donors (p=0.062 vs. the group O patients who did demonstrate ABO antibodies.) Although the p values do not reach conventional significance, the results suggest that it is best to manage patients with ABO antibodies with platelets from group O donors. This analysis left us with 20 SPRCA-positive/LCT-negative cases which could not be explained by ABO antibodies. This represents only 7 percent of the total population. We are in the process of trying to determine whether these sera have HLA antibody which can not be detected by lymphocytotoxicity or antibody to platelet-specific antigens. Although this is a very interesting area, such patients represent only a very small minority among the total patient population needing platelet support. Thus the previously expressed concern that the LCT assay may miss antibody to platelet-specific antigen does not appear to be a significant practical problem in the vast majority of patients refractory to platelet transfusions. Management of the Alloimmunized Patient The classic HLA approach began with the paper of Yankee et al17 who compared the use of platelets from HLA-identical siblings with randomly selected platelets in highly-immunized patients. There was no response to randomly selected products but excellent response to perfect HLA matches. Since it is generally not realistic to support patients with platelets from siblings, HLA matches have subsequently been chosen from the general population. Duquesnoy et al14 described what can be called the classic ABCD or CREG method. A matches are complete four-antigen matches. BU matches are matches in which the donor has no antigen that the recipient lacks although the donor may have an "unknown" (U) antigen. With modern HLA typing, such donors are those who are homozygous for one or two HLA antigens. A BX match contains one or more antigens in the CREGs of the patient which are not present on the patient’s cells. C or D matches contain one or two mismatches in which the mismatched antigens are not even from one of the recipient’s CREGs. Duquesnoy et al studied responses of various levels of matches in patients who were refractory to randomly selected products. There was "good news" in that many patients got good responses to platelets matched by these criteria. The "bad news" was that the predictive power of the method was limited because there were poor responses to many BX matches and good responses to many C and D matches. We can understand this poor predictive capacity (and, of course, poor clinical responses) based on the serologic results that we have just described. Many of these patients were presumably lightly immunized with, perhaps, antibody to only one CREG. Thus, some C and D matches would be successful in them. On the other hand, patients with intra-CREG antibody would be expected to respond poorly to some BX matches. In fact, Dahlke and Weiss18 described specific BX matches which were likely to fail presumably because of the presence of such intra-CREG antibody. Furthermore, in practice in the real world, there is often a considerable period of time between identifying a good potential donor and actually having a transfusable product in hand because of the time required for recruitment, pheresis, and infectious disease testing. Therefore, this approach needs to be supplemented with one or more forms of serologic assessment which will identify a potentially compatible product among those which are immediately available. Some investigators have simply performed SPRCA crossmatching against randomly selected apheresis products. Rachel et al19 first described generally poor responses to crossmatch-incompatible products and generally good responses to crossmatch-compatible products, although a significant fraction of crossmatch-compatible products gave poor CCIs. In a more recent publication, Gelb and Levitt20 reported a mean CCI of 1,800 for randomly selected products and 9,800 for crossmatch-compatible products. Overall, 59 percent of crossmatched products produced CCI’s greater than 7,500. However, the Gelb and Levitt data have to be put into perspective. There were no compatible units found for 10 of 76 patients because of very high degrees of alloimmunization. Furthermore, in the other 66 patients, the mean percentage of screened units that were compatible was 69 percent (range 24-100) suggesting that many of the patients who were actually transfused were not severely immunized. One draws the conclusion that this method is probably satisfactory for mild and moderate degrees of alloimmunization but less helpful for the severely immunized. Furthermore, the method may find a potentially useful product incompatible. As mentioned above, 66 percent of group O patients have ABO antibodies demonstrated by SPRCA. Such an antibody would identify a perfect HLA match as incompatible in crossmatching. Another approach is the antibody specificity prediction (ASP) method recently described by Petz et al21. In brief, they used the LCT assay to identify the antigens to which the patient had formed antibodies and then provided the patient with antigen-negative platelets, i.e., platelets which lack the antigens to which the patient has antibody. In their report, they describe this method as being nearly equivalent to providing A and BU matches and superior to random crossmatching. In principle, one does not even have to know the patient’s HLA type. It is frequently possible to find an antigen-negative product in inventory for immediate distribution to the hospital. At our center, our experience has led us to an approach which combines all of these methods. We have approximately 30-50 products in inventory every day, 90 percent of which have been obtained from donors who have previously been HLA typed. Many products are simply distributed from this inventory. Over a 6-month period we analyzed 1,161 products which had been sent to 93 patients at 33 hospitals. We were fortunate to be able to get CCIs back on 70 percent of them. Two-thirds of the products were selected strictly by CREG criteria supplemented with antigen-negative products identified by the LCT assay. In one-third, crossmatching was also used. We generally reserve crossmatching for patients who are not doing well with products distributed according to HLA criteria. We analyzed the percentage of transfusions which produced CCIs greater than 7,500 in patients on whom we had at least three CCIs. In general, if the PRA was less than 70 percent, good transfusion responses were seen in approximately 80 percent of cases for products that were distributed simply by HLA criteria. However, when the PRA was greater than 70 percent, only about 25 percent of the patients did well. One needs to emphasize that this data includes all patients, including those needing platelets on an emergency basis to which we responded before antibody screening results became available. For the products which were also crossmatched, the patients with PRAs less than 80 percent did well, but, even with crossmatching, the majority of patients with PRAs in the range 80-100 percent failed even with crossmatch-compatible products, suggesting that crossmatching misses significant antibodies when the level of immunization is very high. We have tried to analyze these failures of crossmatch-compatible products. We looked at all such products in which a pre- and post-count was reported to us and in which we also had an LCT antibody screen done by our laboratory within the month prior to transfusion. Antigen-positive transfusions (those in which the donor had antigen to which the patient had antibody) produced CCIs greater than 7,500 in only 9 of 46 transfusions (20%). On the other hand, antigen-negative products (those in which the donor had no antigen to which the patient had antibody) produced CCIs greater than 7,500 in 68 of 140 transfusions (49%)(p < 0.001). This result suggested that, even though the LCT and SPRCA assays produce results which are concordant overall in terms of PRA, crossmatching may miss clinically significant, individual HLA antibodies which can be detected by the LCT assay. These results raise the question as to whether crossmatching provides any "value added" over providing antigen-negative products. In this same group of patients, we examined 145 crossmatched products in which an LCT screen had been done for the patient in the preceding month and that LCT had identified no antibody to antigen that was on the platelets that were being crossmatched i.e., these were all antigen-negative products. Nonetheless, fully a third were incompatible by crossmatching (47 of 145). Of these 47, 22 (47%) were in group O patients and the products being crossmatched were group A or B. Presumably, crossmatching was picking up ABO antibody in many of these antigen-negative products. On the other hand, this left us with 25 (17%) of the 145 antigen-negative products which were incompatible but not on the basis of ABO antibody. Again, we are not sure what it is that crossmatching is picking up in such cases, perhaps HLA antibody not detected by the LCT screen, or platelet-specific antibody. Furthermore, since we do not distribute crossmatch-incompatible products for transfusion, we do not know the clinical relevance of these serologic reactions. However, the results led us to believe that both the ASP method and the crossmatching method should be used together in highly immunized patients. In effect, we are essentially recommending an approach which is identical to what we do for red cell transfusion. We first identify the antigens to which the patient has antibody and then choose products for crossmatching from among antigen-negative products. Summary Our experience and that reported in the literature suggests to us an approach to the refractory patient which is summarized in Table I. We recommend that all patients have a serologic study, either an LCT or SPRCA screen. If no antibody can be identified, one can suggest to the hospital that the patient might do just as well with randomly-selected products. The only caveat is that the patient may be very early in his or her alloimmunization so that it is wise to check another antibody screen in 2-3 weeks. If the antibody screen suggests ABO antibody in a group O patient, one can recommend the use of platelets from group O donors. If one is dealing with the relatively rare patient who is SPRCA-positive/LCT-negative and the SPRCA does not suggest an ABO antibody, one can manage the patient with crossmatching and send serum to a specialized laboratory which can perform investigations looking for platelet-specific antibody. If the PRA is in the range of 1-70 percent because of the presence of HLA antibody, several approaches probably are effective. One can either simply crossmatch randomly selected products or provide from inventory products which lack the antigen to which the patient has antibody. The patient should be monitored with CCIs to be certain that a good response is being achieved. Patients with PRAs greater than 70 percent are more complex to deal with. We believe that crossmatching alone frequently leads to inferior results, in part because one may crossmatch 10-30 randomly-selected products and find none which is compatible. Therefore, we recommend that one crossmatch using antigen-negative platelets. Furthermore, we recommend the selective recruitment of A and BU matches from an HLA-typed donor pool because, as the PRA approaches 100 percent, it is commonly only such perfect or nearly perfect matches that will provide adequate support. We generally ask the hospital to place the patient on a transfusion schedule if such selective recruitment of donors is to be performed. In these cost-conscious times, many hospitals do not wish to obtain such serologic assessments and pre- and posttransfusion counts. Actually, the cost of platelet counting and a once-monthly antibody screen is small relative to the cost of an ineffective apheresis product. Furthermore, the hospital may not wish to put the patient on a transfusion schedule since they may receive a product which the clinician does not wish to use. Nonetheless, we believe that optimal support of these patients makes all three of these relatively modest expenditures highly desirable. Finally, many clinical services request not only a matched product but also a CMV-negative product. Only 40 percent of our donors are CMV-negative so that such a request reduces the size of the donor pool and our ability to provide an optimal product. We suggest to the hospital that it may be in the best interest of the patient to use leukoreduction as a method for making the product CMV safe and thus allow the patient to receive an optimally matched product. Acknowledgements The author thanks Drs. Alfred Katz, Alfred Grindon, and S. Gerald Sandler for their careful review of this manuscript.
Uploaded: Nov 30, 2000 |
