Chronic transfusion in patients with thalassemia is often complicated by red blood cell (RBC) alloantibodies to the lacking antigens on the patients’ RBCs. The prevalence of alloantibodies ranges from 4.25 to 37 percent in patients with thalassemia.1–6 Clinically significant alloantibodies can shorten transfused erythrocyte survival due to hemolytic transfusion reactions. To reduce the alloantibody risk, RBC antigen serotyping for Rh (C, c, E, e) and MNS hybrid glycophorins, especially for MNS7 (Mia), is recommended before the first transfusion by our national thalassemia guidelines. Subsequently, Rh- and Mia-matched blood units are transfused.7 However, phenotyping data before the first transfusion are unavailable at our institute, and most patients with thalassemia have been receiving blood regularly since childhood.
Serologic methods are the standard for determining ABO, D, and common blood group antigens. Nevertheless, this approach in chronically transfused patients with thalassemia is inaccurate because of the contaminating donor erythrocytes in patient circulation. DNA-based typing is superior in situations of recent transfusions.8 Moreover, it can guide the antigen-matched transfusion in individual patients. Furthermore, autoantibodies are frequently found in patients with thalassemia, causing crossmatch problems.9 The prevalence of autoantibodies is 0.47–28.2 percent by direct antiglobulin test and 57.6 percent by flow cytometry depending on study population and transfusion policy.10–13
In this study, we implemented DNA-based typing for chronically transfused patients with thalassemia and provided antigen-matched RBC units accordingly. A better-matched unit should prolong the RBC survival
In our tertiary care hospital with many referred patients with thalassemia, RBC phenotype data before the first transfusion are unavailable for most patients. Therefore, we enrolled this group of patients who previously had crossmatch problems or history of unsatisfactory pre-transfusion Hb after chronic transfusion. Patients who did not receive regular transfusion (
Donor units were antigen matched using hemagglutination in gel cards (Bio-Rad; Cressier, Switzerland) for ABO, Rh (D, C, c, E, e), Mia, M, S, Jka, Jkb, Fya, Fyb, and Dia, since they are the most common clinically significant alloantibodies in Thailand.10,14 Genotyping was not performed on the donor blood components. Typing for Rh (C, c, E, e), Jka, Jkb, Fya, and Fyb was performed according to the manufacturer’s instructions (Bio-Rad). Anti-Mia and anti-M were produced by the Thai National Blood Centre, Thailand. S was assessed using the reagent antibody from CE-IMMUNDIAGNOSTIKA GmbH (Germany). Anti-Dia was produced in-house from patient sera with anti-Dia.
DNA-based typing for common blood group antigens was performed by the BLOODchip ID Core XT kit (Progenika Biopharma, Derio, Spain) and the Luminex genotyping platform (Luminex, Austin, TX), which interrogates 29 polymorphisms to predict the expression of 37 antigens in 10 blood group systems. DNA samples were extracted from EDTA blood by spin-column separation (QIAmp; QIAGEN, Valencia, CA). DNA concentration was determined by measurement of optical density at 260 and 280 nm. Extracted DNA was kept at –20°C until testing. Samples were processed with one positive and one negative control in each run.
The primary outcome was to detect new alloantibodies after transfusion of antigen-matched units, since RBC allo-immunization is associated with a risk of hemolytic transfusion reaction. According to the treatment guideline for chronically transfused patients with thalassemia, a pre-transfusion Hb level between 9 and 10.5 g/dL is recommended to promote growth, improve physical activities, and suppress abnormal bone marrow expansion.15 Therefore, the pre-transfusion Hb response to antigen-matched transfusion was assessed. Marked improvement was defined as an increase of mean pre-transfusion Hb of at least 1 g/dL from 6 months from the patient’s enrollment to the last follow-up visit compared with that of the 12-month period before enrollment.
Crossmatch problems could delay blood transfusion for some patients with thalassemia having an autoantibody or antibody to a high-prevalence antigen even though antigen-matched units for all clinically significant common antibodies were issued. Therefore, the secondary outcome was to assess the response to crossmatch-positive but antigen-matched RBC units. Also, finding antigen-matched units for all clinically significant common antibodies was both time and labor consuming. We attempted to identify the predictors of patient responses to antigen-matched RBC units, such as history of three or more antibodies and patient age.
Patients received regular transfusions every 1–4 weeks. Laboratory tests including complete blood count (CBC) and ferritin were performed every 3 months. Therefore, pre-transfusion Hb was not available at every time point. The safety outcome was the incidence of hemolytic transfusion reactions. Definition of hemolytic transfusion reaction was based on the National Healthcare Safety Network Biovigilance Component Hemovigilance Module Surveillance Protocol by the U.S. Centers for Disease Control and Prevention.16 According to our hospital hemovigilance policy, all transfusion reactions were reported to the Transfusion Medicine Unit.
Continuous variables were summarized descriptively by means, medians, ranges, and standard deviations (SDs) when appropriate. The Wilcoxon signed-rank test was used to compare the baseline pre-transfusion Hb and mean pretransfusion Hb levels from 6 months after the enrollment. The Mann-Whitney
Statistical analysis of the relationship between predictors and the secondary outcome of the patients used bivariate comparisons of categorical variables by two-tailed χ2 tests. A
Twenty-four transfusion-dependent patients with thalassemia were initially enrolled; two were later excluded because of lack of transfusion. Nineteen (86.4%, 19/22) were female. The median age was 24 years (range 2–58 years). Thirteen (59.1%, 13/22) patients were diagnosed with beta-thalassemia Hb E. Eight patients (36.4%, 8/22) were homozygous for beta-thalassemia, and one patient (4.5%, 1/22) had Hb H Constant Spring. Twelve patients had three or more alloantibodies, five had two alloantibodies, four had a single alloantibody, and three had no alloantibody. The three most common antibodies were anti-Mia, anti-E, and anti-c (Table 1). All patients had no previous history of hemolytic transfusion reactions or hyperhemolytic transfusion events.
Clinical characteristics of 24 patients with thalassemia
Patient number | Age, years | Gender | Type of thalassemia | Splenectomy | Complications/comorbidities | Antibody specifcity | Predicted phenotype* | Mean pre-transfusion Hb levels before antigen-matched transfusion (g/dL) | Mean pre-transfusion Hb levels after antigen-matched transfusion (g/dL) |
---|---|---|---|---|---|---|---|---|---|
1 | 29 | F | Beta-thalassemia Hb E | No | Iron overload | Lea, Leb, E, Mia, Fyb, Dia | E–, K–, Fy(b–), S–, Mi(a–), Di (a–) | 4.5 | 5.5 |
2 | 12 | M | Homozygous beta-thalassemia | Yes (3 years before the study) | Iron overload, chronic portal, and splenic vein | Mia, Leb, Jkb, unidentifed | E–, K–, Fy(b–), Jk(b–), Mi(a–), Di (a–) | 7.2 | 9.1 |
3 | 3 | F | Homozygous beta-thalassemia | No | Iron overload | E, c, Mia, S | E–, c–, K–, Fy(b–), N–, S–, Mi(a–), Di (a–) | 8.9 | 8.6 |
4 | 36 | F | Beta-thalassemia Hb E | No | Iron overload | E, c, Mia, S, Jkb | E–, c–, K–, Jk(b–), Fy(b–), N–, S–, Mi(a–), Di (a–) | 6.8 | 7.2 |
5 | 40 | F | Homozygous beta-thalassemia | No | Iron overload | M, c, K, Jkb, Mia | E–, c–, K–, Jk(b–), M–, S–, Mi(a–), Di (a–) | 7.3 | 6.9 |
6 | 39 | F | Beta-thalassemia Hb E | Yes (23 years before the study) | Iron overload | Lea, Leb, S, unidentifed | K–, Fy(b–), N–, S–, Di (a–) | 6.9 | 7.4 |
7 | 14 | M | Beta-thalassemia Hb E | No | Iron overload | E, c, Mia, Dia | E–, c–, K–, Fy(b–), N–, S–, Mi(a–), Di (a–) | 7.9 | 7.7 |
8 | 20 | F | Homozygous beta-thalassemia | Yes (16 years before the study) | Iron overload, transaminitis | E, c, S, Chido | E–, c–, K–, N–, S–, Di (a–) | 8.7 | 7.4 |
9 | 27 | M | Homozygous beta-thalassemia | Yes (23 years before the study) | Iron overload | E, c, Lea, Fyb | E-, c-, K-, Fy(b-), S-, Mi(a-), Di(a-) | 4.7 | 3.6 |
10 | 42 | F | Beta-thalassemia Hb E | No | Iron overload | Jka, Mia, unidentifed | C–, K–, Fy(b–), Jk(a–), M–, S–, Mi(a–), Di (a–) | 5.8 | 6.1 |
11 | 53 | F | Hb H Constant Spring | Yes (20 years before the study) | Iron overload, pulmonary hypertension, chronic deep vein thrombosis, and pulmonary embolism | E, c, Mia | E–, c–, K–, S–, Mi(a–), Di (a–) | 9.8 | 9.7 |
12 | 44 | M | Homozygous beta-thalassemia | Yes (during the study) | Iron overload, extramedullary hematopoiesis at spine causing neurodefciency | E, c, Mia | E–, c–, K–, Fy(b–), Mi(a–), Di (a–) | 6.3 | 7.3 |
13 | 58 | F | Beta-thalassemia Hb E | Yes (13 years before the study) | Iron overload | E, auto-C | E–, c–, K–, Jk(b–), Fy(b–), S–, Mi(a–), Di (a–) | 8.1 | 7.6 |
14 | 20 | F | Homozygous beta-thalassemia | Yes (6 years before the study) | Iron overload, pulmonary hypertension | Mia, S | E-, c-, K-, Jk(b-), S-, Mi(a-) | 6.5 | 7.3 |
15 | 20 | F | Beta-thalassemia Hb E | No | Iron overload, epilepsy | E, c | E–, c–, K–, S–, Mi(a–), Di (a–) | 3.7 | 5.1 |
16 | 2 | F | Homozygous beta-thalassemia | No | None | Jka, unidentifed | K–, Jk(a–), N–, S–, Mi(a–), Di (a–) | 9.0 | 9.2 |
17 | 19 | F | Beta-thalassemia Hb E | No | Iron overload | E, unidentifed | E–, c–, K–, Jk(b–), Fy(b–), M–, S–, Mi(a–), Di (a–) | 7.9 | 8.3 |
18 | 7 | F | Beta-thalassemia Hb E | No | Iron overload | Chido | E–, c–, K–, Jk(b–), Fy(b–), S–, Mi(a–), Di (a–) | 8.7 | 9.1 |
19 | 11 | F | Beta-thalassemia Hb E | No | Iron overload, ventricular septal defect/atrial septal defect | Mia | E–, K–, N–, S–, Mi(a–), Di (a–) | 7.6 | 7.3 |
20 | 35 | M | Beta-thalassemia Hb E | Yes (22 years before the study) | Iron overload | E | E-, c-, K-, Fy(b-), S-, Mi(a-), Di(a-) | 3.7 | 4.9 |
21 | 28 | F | Beta-thalassemia Hb E | Yes (23 years before the study) | Iron overload | Nonspecifc cold | E-, K-, Fy(b-), N-, S-, Mi(a-), Di(a-) | 8.5 | 8.9 |
22 | 6 | F | Homozygous beta-thalassemia | No | Iron overload | No alloantibody | E–, c–, K–, Jk(a–), N–, S–, Mi(a–), Di (a–) | Patient did not receive regular transfusion before enrollment; excluded from study. | |
23 | 17 | F | Beta-thalassemia Hb E | No | Iron overload | No alloantibody | E–, c–, K–, Fy(b–), N–, S–, Mi(a–), Di (a–) | 9.2 | 7.6 |
24 | 27 | F | Beta-thalassemia Hb E | No | Iron overload | No alloantibody | E–, c–, K–, Jk(b–), N–, S–, Mi(a–), Di (a–) | Patient did not receive regular transfusion before enrollment; excluded from study. |
Red blood cell phenotype predicted by genotyping method. Hb = hemoglobin.
After the enrollment, there were 717 transfusion episodes comprising 1209 antigen-matched units. Blood transfusion to each patient ranged from 22 to 119 units. During the median follow-up of 25.5 months (range 10–34 months), no new alloantibodies were detected.
Among the 24 patients, 2 patients (patients 22 and 24) had not received regular transfusions before the enrollment and were excluded. For the remaining 22 patients, there was no statistically significant difference between the pre-transfusion Hb level before date of enrollment and the level after 6 months of antigen-matched transfusions from date of enrollment until the last visit (
Interestingly, five patients (patients 1, 2, 12, 15, and 20) showed marked improvement in mean pre-transfusion Hb levels (≥1.0 g/dL) compared with levels before the enrollment. Of these patients, three patients had three or more alloantibodies, one patient had two alloantibodies, and one patient had one alloantibody.
Patient 1, with beta-thalassemia Hb E, had multiple antibodies. Her average baseline pre-transfusion Hb was 4.5 g/dL. She received antigen-matched transfusions for 26 months. Six months after her enrollment, her mean pre-transfusion Hb was 5.5 g/dL.
Patient 2, with homozygous beta-thalassemia, had multiple antibodies and did not respond well to transfusion. Three years previously, he underwent a splenectomy in an attempt to maintain better pre-transfusion Hb. Nevertheless, his pre-transfusion Hb was unimproved after splenectomy. His baseline pre-transfusion Hb after splenectomy was 7.2 g/dL. He received antigen-matched transfusions for 28 months. Six months after the enrollment, his mean pre-transfusion Hb was 9.1 g/dL.
Patient 12, with homozygous beta-thalassemia, had anti-E, anti-c, and anti-Mia. His average baseline pre-transfusion Hb was 6.3 g/dL. He received antigen-matched transfusions for 14 months. Six months after his enrollment, his mean pre-transfusion Hb was 7.3 g/dL. Note that he underwent a splenectomy 8 months after the enrollment, however. Seven months after splenectomy, his Hb was 9.0 g/dL, and he has not received a blood transfusion since then. Thus, the improvement of his pre-transfusion Hb could be a result of the splenectomy.
Patient 15, with beta-thalassemia Hb E, had anti-E and anti-c. Her average baseline pre-transfusion Hb was 3.7 g/dL. She received antigen-matched transfusions for 34 months. Six months after this enrollment, her mean pre-transfusion Hb was 5.1 g/dL.
Patient 20, with beta-thalassemia Hb E, had anti-E. Antigen-negative RBC units for E, c, and Mia were given, but his pre-transfusion Hb was unimproved. His baseline pretransfusion Hb was 3.7 g/dL. He received antigen-matched transfusions for 28 months. Six months after this enrollment, his mean pre-transfusion Hb was 4.9 g/dL.
The mean timing between transfusions before and after antigen-matched transfusion was not statistically different (
Although antigen-matched units for Rh (D, C, c, E, e), Mia, M, S, Jka, Jkb, Fya, Fyb, and Dia were issued for all patients, six patients (27.3%, 6/22; patients 2, 4, 8, 9, 14, and 18) showed positive crossmatch reactions with some units. Five of them had autoantibodies, and the other had anti-Chido; all resulted in positive crossmatches. In these six patients, there were 210 transfusion episodes, 74 of which (35.2%) showed positive crossmatch results. The grades of the crossmatch reactivity (
Before this study was performed, autoantibody or antibody to a high-prevalence antigen often complicated the pre-transfusion process and delayed the transfusion schedule. During this study, antigen-matched units for Rh (D, C, c, E, e), Mia, M, S, Jka, Jkb, Fya, Fyb, and Dia could be prepared in advance; therefore, the patients received transfusions on time.
For the patient with anti-Chido, Chido– RBCs were not provided because it is not a clinically significant antibody. For the patients with autoantibodies, antigen-matched units were found ahead of the transfusion date and the least incompatible antigen-matched unit(s) were selected for transfusion on the day of transfusion.
In the 22 patients with negative versus positive cross-match results, the responses to antigen-matched transfusions were not statistically different (
We hypothesized that patients with three or more antibodies would respond well to antigen-matched transfusions. However, the response to antigen-matched transfusions between those patients with versus those patients without three or more antibodies was not statistically significant among the 22 patients (
This study determined that the age of the patient before this enrollment could not be a predictor for successful antigen-matched transfusions (
To the best of our knowledge, this is the first report of clinical outcomes of phenotype-matched transfusions to multiply transfused patients with thalassemia guided by patient DNA-based typing. Of the 22 patients without serologic phenotypes, data before the first transfusion were included, and no new alloantibodies were detected. Hb levels of five patients were increased, and one patient required less frequent transfusions after receiving antigen-matched transfusions, suggesting a prolongation of transfused erythrocyte survival (5 of 22 evaluable patients, or 22.7% response rate), although there was no significant change for the entire group. The fifth patient with Hb improvement also underwent a splenectomy and, therefore, the effect of this protocol cannot be determined.
Successful maintenance of pre-transfusion Hb in chronically transfused patients with thalassemia is an important goal for better quality of life, lower frequency of transfusion, lower serum ferritin, and fewer transfusion reactions.15,17 RBC alloimmunization is a challenge when providing compatible blood units. Although many factors contribute to this undesirable event, the most important factor is the exposure to foreign RBC antigens.18,19 Therefore, antigen-matched transfusion is a goal for success. Additionally, the efficacy of transfusion for chronically transfused patients with thalassemia could be affected by the timing between transfusions and host factors, such as splenectomy and comorbidities, as well as the presence of precipitating factors, such as infections.
Because the baseline serologic phenotype of most patients is unavailable, a report of regularly transfused Thai patients with thalassemia showed that the presence of circulating donor RBCs leads to mixed-field reactivity and inaccurate results.20 Discrepancies between serologic phenotypes performed by different laboratory personnel at different time points of testing were also observed. Therefore, DNA-based typing offers an advantage over traditional serologic typing to accurately predict the identity of the patient’s RBC antigens after recent transfusion. Moreover, DNA-based typing may provide benefits to patients with autoantibodies, multiple alloantibodies, antibodies to high-prevalence antigens, or antibodies against antigens with unavailable antisera.8
The extended molecular-matched RBC transfusion for chronically transfused patients can improve the effectiveness of transfusion by decreasing the alloimmunization rate.17,21 Despite these benefits, this approach is currently not widely available in Thailand because of the high cost for genotyping a large inventory of donor RBCs. Therefore, DNA-based typing is only performed for patients, and phenotype-matched units are provided according to patient genotyping data.
Finding compatible blood units for patients with multiple alloantibodies is laborious and time-consuming. After protocol implementation, antigen-matched units were procured before the planned patient visits. Although some crossmatches revealed serologic incompatibility due to autoantibody or clinically insignificant antibody to a high-prevalence antigen, blood units were issued on time because all clinically significant antigens were matched. The least incompatible antigen-matched units from the donor inventory would be transfused to the patient, and the antibody identification retested on every visit. The transfusion of incompatible RBC units is not a standard practice. However, it is not possible in some patients with autoantibody or clinically insignificant antibody to high-prevalence antigen to provide compatible units. In this case, administration of antigen-negative RBC units is the recommended approach. Interestingly, hemolytic transfusion reactions were not detected in our patients under this protocol.
Prophylactic antigen-matching has been shown to reduce new antibody formation.11,22 A recent study in patients with sickle cell disease in the United States showed new antibody formation for the unmatched antigens.21 We matched only common clinically significant RBC antigens in Thailand, but no new alloantibodies were detected. One explanation is the ethnic differences between donors and patients in the United States.23 Sickle cell disease affects mainly African Americans, whose blood donation rates are only 25–50 percent of those of white individuals.24 In Thailand, blood donors share the same ethnicity with the patients. Therefore, there is a better chance to obtain phenotype-matched blood units for Thai patients with thalassemia.
The predictors of good responses to antigen-matched blood could not be identified. The number of patients in our study was probably too small to find weak predictors. Therefore,
To apply this study into our routine work, DNA-based typing should be performed for chronically transfused patients with thalassemia who have no available phenotyping data. The extended antigen-matched transfusion could apply to patients with multiple antibodies, autoantibody, or a clinically insignificant antibody to a high-prevalence antigen but not for patients without antibody. This practice will help the majority of patients with difficult pre-transfusion testing to receive blood on schedule.
In conclusion, DNA-based typing is helpful in transfusion management of chronically transfused patients with thalassemia, especially when phenotype data are unavailable. The genotype data allow us to provide antigen-matched units despite a positive crossmatch. In this study, no new alloantibodies were detected after the median follow up of 25.5 months and average number of 50 units/patient. A subgroup of patients showed increases in pre-transfusion Hb. A larger prospective clinical trial is needed to evaluate the definitive role of DNA-based typing in transfusion-dependent patients with thalassemia.