The majority of weak D phenotypes result

Transcription

1 BRIEF REPORT How do we identify RHD variants using a practical molecular approach? Carine Prisco Arnoni, 1 Flavia Roche Moreira Latini, 1 Janaína Guilhem Muniz, 1 Diana Gazito, 1 Rosangela de Medeiros Person, 1 Tatiane Aparecida de Paula Vendrame, 1 José Augusto Barreto, 1 and Lilian Castilho 2 Serologic resolution of Rh discrepancies due to partial D or weak D phenotypes is a frequent problem encountered during routine typing that can be solved by RHD genotyping because it provides better characterization of these variants. The objective of the current study was to develop algorithms for identification of D variants in multiethnic populations based on a logic sequence of molecular tests using a large number of atypical RhD specimens. Thus, a total of 360 blood samples with atypical D antigen expression were analyzed. A previously published multiplex polymerase chain reaction (PCR) procedure was performed and depending on multiplex PCR analysis, the associated RHCE allele, and D variant frequency in our population, an algorithm was developed composed of six flow charts using specific PCR restriction fragment length polymorphism and/or specific exon sequencing. This strategy allowed the identification of 22 different variants with few assays and a much reduced cost. This study describes a simple and practical algorithm that we use to determine RHD genotypes in samples with unknown RHD. This strategy is relatively easy to implement and the algorithm can be adapted to populations with various ethnic backgrounds after an initial assessment of the type and frequency of D variants. Essentially, we demonstrate that sequencing of all RHD exons is not necessary for the identification of the majority of known D variants. The majority of weak D phenotypes result from single-nucleotide polymorphisms (SNPs) in RHD encoding amino acid changes within either the membrane-spanning domains or the cytoplasmic loops of the protein. These changes can interfere with the integration of the RhD protein in the membrane leading to a reduced number of D antigen sites on red blood cells (RBCs). 1 Partial D, in contrast to weak D, is characterized by amino acid changes outside of the membrane that can alter or create new epitopes. 2 Therefore, individuals with partial D can make anti-d when stimulated by transfusion or pregnancy. In fact, many partial D are typed as D+ by direct agglutination and these individuals will be identified only after anti-d formation. 3 RHD genotyping is useful for precise characterization of partial D and weak D types in donors and recipients and is a clinically important approach to prevent alloimmunization of recipients with partial D, usually typed as D+, when exposed to D+ RBCs or to some of the weak D RBCs typed as D by serologic methods. However, RHD genotyping is not easy, because it is influenced by the size of the gene, by the presence of rearrangements with RHCE, 3 and by the high number of SNPs that are recognized in variants found among different ethnic groups. 4 A multitude of different primers and probes is available for molecular genotyping of RHD making difficult the ABBREVIATION: SNP(s) = single-nucleotide polymorphism(s). From the 1 Colsan Associação Beneficente de Coleta de Sangue, São Paulo, SP, Brazil; and 2 INCTs Hemocentro-Unicamp, Campinas, SP, Brazil. Address reprint requests to: Carine Prisco Arnoni, Colsan Associação Beneficente de Coleta de Sangue, Avenida Jandira 1260, Indianópolis CEP , Brazil; carine.arnoni@colsan.org.br. Received for publication July 26, 2013; revision received November 28, 2013, and accepted November 29, doi: /trf AABB TRANSFUSION **;**:**-**. Volume **, ** ** TRANSFUSION 1

2 ARNONI ET AL. choice of a testing strategy that is often complex and expensive and sometimes results in waste of time, resources and reagents. This complexity led us to search for an easy and rational strategy for identification of the most common D variants using a shorter and less expensive path. Our algorithm is based on the molecular genetics of RHD, D polymorphisms, frequencies of D variants in different populations, and their association with RHCE alleles using public databases as reference, including the variants recognized by ISBT 5 and the tools used to classify variants available on the Rhesus Site. 6 All variants that we detected are associated with RHCE alleles already reported in the literature, 6 except for DV Type 2 that in our population is associated with an RHCE*ce. This algorithm is organized as a sequence of testing protocols based on a previous analysis of 360 genotyped Brazilian donor samples. Essentially, 84 samples (23%) were solved using the multiplex polymerase chain reaction (PCR) and one PCR restriction fragment length polymorphism (RFLP), and 135 samples (37%) were solved using the multiplex PCR and two PCR-RFLP. Consequently, we characterized 60% of samples with altered D antigen expression with few tests and without sequencing. Additionally, 121 samples (33%) were solved using the multiplex PCR, up to three PCR-RFLP, and the sequencing of one exon. Frequency of D variants found in our population and associated RHCE alleles are shown in Table 1. RATIONALE OF THE MOLECULAR APPROACH The algorithm was based on the analysis of 360 samples from blood donors with atypical D expression, including discrepancies or reactivity weaker than 3+ in D typing. The first step in the strategy that we chose was the performance of a multiplex PCR for the identification of RHD variants as described by Maaskant-van Wijk and colleagues. 7 Depending on the results of the RHD multiplex PCR analysis, the associated RHCE alleles and the D variant frequencies found in our population, we directed our search with basis on SNPs previously described for the most common D variants (ISBT 5 and Rhesus Site 6 ) through specific PCR-RFLP and/or exon-specific sequencing. 8 Following this path we were able to systematize the investigation, find the polymorphisms, and consequently, distinguish the variants, resulting in flow charts from 1 to 6 (Figs. 1-6). We initially analyzed the sequences and positions of the primers designed by Maaskant-van Wijk and colleagues 7 to amplify RHD Exons 3, 4, 5, 7, and 9 in the multiplex PCR and noted that they were located in regions where there are SNPs involved with several D variants (Table 2). Based on multiplex PCR result, we could rule out some variants and investigate others, and together with the Rh-associated RHCE alleles and frequency of variants, we could decide which protocol should be applied next. It TABLE 1. Frequency of D variants found in a population of 360 Brazilian blood donors, RHCE allele association, and flow chart used for identification of the variant Associated RHCE allele Flow chart D variant Number Percent ce 3 Weak D Type ce 3 Weak D Type Ce 1 Weak D Type ce 2 Weak D Type Ce 1 Weak D Type Ce 1 Weak D Type ce 3 DAR Ce 1 D VII ce 1 DAU ce 1 DMH ce 1 DAU ce 4 DOL ce 4 DAU ce 1 DAU Ce 1 DNB Ce 1 Weak D Type ce 4 D V Type ce 5 D III Type ce 6 D IVa Type ce 3 Weak D Type ce/ce 4 DV Type 1/RHDψ ce/ce 3 DAR1/Weak D Type ce/ce 3 Weak D Type 4.2.2/Weak D Type Total TRANSFUSION Volume **, ** **

3 STRATEGY TO IDENTIFY D VARIANTS Fig. 1. Flow Chart 1 applied when all exons of the multiplex PCR procedures are amplified. sample comes from an individual who is 455C. The 455A>C substitution is present in several D variants, including some types of DIII, DVI, DIVa, and hybrid genes (RHD-CE-D). Whenever the Exon 3 is not amplified in the multiplex PCR, we search for these variants taking into account the associated RHCE alleles. It is important to note that the presence of amplification products of Exon 3 in the multiplex PCR excludes the aforementioned variants. The same logic is applied to the other exons analyzed in the multiplex PCR. Fig. 2. Flow Chart 2 performed when only Exon 9 is not amplified in the multiplex PCR. is important to note that some exceptions related to the inheritance of Rh haplotypes are possible. The following is an example of the analysis of Exon 3 based on primer location, SNPs, and the genetic feature of D variants. The Exon 3 antisense primer ( nucleotides of RHD gene) is located in the region where some of the variants have a cytosine instead of an adenine at Position 455. Amplification of this exon only occurs when the specific primers recognize an adenine at this position; this product is not amplified when the Multiplex PCR The multiplex PCR approach used was adapted from the method described by Maaskant-van Wijk and coworkers 7 using five RHD-specific primer sets designed to amplify only RHD Exons 3, 4, 5, 7, and 9. PCR was performed as previously described 7 and the results were analyzed in agarose gel, as shown in Fig. 7. PCR-RFLP We used restriction digestion (PCR-RFLP) to detect the polymorphisms related to weak D Type 1 (809T>G; ApaLI), weak DType 2 (1154G>C; AluI), weak DType 3 (8C>G; SacI), weak D Type 4.1 (48G>C; ApaI), weak D Type (744C>T; BtsIMutI), DAR-1 (957G>A; BseYI), DAU (1136C>T; NlaIII/ 998G>A; AcuI/835G>A; NlaIII), and DIIIa (186G>T; BstXI). Volume **, ** ** TRANSFUSION 3

4 ARNONI ET AL. All primers were supplied by Sigma- Aldrich Corp. (Woodlands, TX) and restriction enzymes by New England Biolabs (Ipswich, MA). Fig. 3. Flow Chart 3 used to analyze samples when Exons 5 and 4 are not amplified in multiplex PCR. DNA sequencing DNA sequencing was performed according to standard protocols. Primers used were previously reported. 8 PCR products were purified with Exonuclease I and thermosensitive alkaline phosphatase (FastAP, Fermentas, Hanover, MD) and purified PCR products were submitted to sequencing reaction with a cycle sequencing kit (BigDye Terminator, Version 3.1, Applied Biosystems, Foster City, CA). The product was purified with a purification kit (BigDye X-terminator, Applied Biosystems) and sequencing analysis was performed on a genetic analyzer (3500xL, Applied Biosystems). Electropherograms were analyzed using Sequencing Analysis Software (Applied Biosystems). Identification of the RhD variants The final identification of the RhD variant is based on the analysis of both, exons that were not amplified and exons that were amplified (Table 3). Table 2 lists the SNPs associated with each exon and the potential variants that are excluded when individual exons are not amplified in the multiplex PCR. Using this approach, we developed algorithms for the identification of D variants that are described in the six flow charts, according to the results of the multiplex PCR (Figs. 1-6). Table 3 lists the possible results of the multiplex PCR as they lead the investigator to each of the six flow charts, including the SNPs that are present or absent and the most common possible variants. Fig. 4. Flow Chart 4 applied when only Exon 5 is not amplified in multiplex PCR. Flow Chart 1 Figure 1 describes the strategy used when all exons are amplified in the multiplex PCR, which means that no variants related to the SNPs 455A>C, 514C>T or 602C>G, 667T>G, 992A> T, 1048G>C, 1154G>C, and 1193A>T are present. Consequently, we exclude 4 TRANSFUSION Volume **, ** **

5 STRATEGY TO IDENTIFY D VARIANTS and 9 are amplified, we know that SNPs related to them are absent. In the presence of the ce phenotype, we drive the search to weak D Type 4 cluster and in the presence of C or E antigens, we investigate types of DVI. Fig. 5. Flow Chart 5 used when only Exons 7 and 9 are amplified in multiplex PCR. Flow Chart 4 Figure 4 summarizes the steps taken when only Exon 5 is not amplified in the multiplex PCR. The first step, independently of the RHCE alleles, is the investigation of the SNP 697G>C. When we find a guanine at Nucleotide 697, we focus our search to identify DOL1, DOL2, and DOL3, but if cytosine is present at Nucleotide 697, we investigate DV, DBS, DFV, DCS, DTO, and DAU5. When the RHDψ is together with a variant that has the 667T>G SNP, only Exon 5 will not be amplified in the multiplex PCR and Flow Chart 4 will be used. Using this flow chart, we can easily identify the presence of the RHDψ.Itis important to note that heterozygous variants with the RHDψ need to be further investigated and the use of other flow charts may be required. Fig. 6. Flow Chart 6 used when Exons 3 and 7 are not amplified in multiplex PCR. those variants described in Table 2 and conduct search based on associated RHCE allele and on the frequency of the most common variants in our population. In the presence of the C antigen, we focus our search on some European D cluster. When the sample is C and E negative we conduct the search through DAU cluster, searching for the SNP 1136C>T. Flow Chart 2 Figure 2 describes the strategy used when all except Exon 9 are amplified in the multiplex PCR. In this case, we know that SNP 1154G>C and/or 1193A>T are present and we investigate weak D Type 2 in the presence of E antigen (SNP1154G>C) and DIV in the absence of E antigen (SNP 1193A>T). Flow Chart 3 Figure 3 contemplates the most frequent D variants found in our population. When Exons 4 and 5 are not amplified in the multiplex PCR, the SNPs 514C>T or 602C>G and 667T>G might be present. Additionally, when Exons 3, 7, Flow Chart 5 Figure 5 shows a low-frequency profile associated with amplification of Exons 7 and 9 in the multiplex PCR. It means that SNPs 455A>C, 514C>T or 602C>G, and 667T>G, which are related to the absence of Exons 3, 4, and 5, are present, leading to the investigation of some types of DIII and DV. Flow Chart 6 Figure 6 is used when Exons 3 and 7 are not amplified in the multiplex PCR and comprises the sequencing of Exon 3. In this case, only two D variants are possible, DIVa and DIVa Type 2, which are distinguished by the SNP 410C>T. BENEFITS OF THE MOLECULAR APPROACH We herein describe a molecular strategy that allowed a rational approach to identification of D variants in our population. The strategy was based on the identification of key SNPs, the frequency of D variants, and the RHCE alleles associations. Due to the small number of studies showing the prevalence of D variants in our population, the frequencies were determined during this algorithm development. It helped us to define which step should come next to shorten the process. Based on the algorithm described we were able to reduce the number of molecular assays performed and identify D variants in the Brazilian population. It is important to note that the application of this algorithm in other populations probably needs to take Volume **, ** ** TRANSFUSION 5

6 ARNONI ET AL. TABLE 2. Absence of amplification of individual exons in the multiplex PCR 7 suggests the presence of certain SNPs and possible D variants Multiplex PCR: no amplification SNP* Possible D variants Exon 3 455A>C D IIIa,D IIIb,D IIIc,D III Type 4, D III Type 5, D III Type 6, D III Type 8,D IVa,D IVa Type 2, D VI Type 3, D VI Type 4, DKK, DFR Type 5, DHAR, and other hybrid genes. Exon 4 514C>T and/or 602C>G D IIIa,D IIIb,D III Type 5, D III Type 6, D VI Type 1, D VI Type 2, D VI Type 3, D VI Type 4, DAR1, DAR2, weak D 4.0, weak D 4.1, weak D 4.2.2, weak D 4.3, weak D Type 40, weak D Type 14, weak D Type 51, DBU, DFR 1, DFR 2, DFR 3, DFR Type 5, DHAR, and other hybrid genes. Exon 5 667T>G D IIIa,D IIIb,D III Type 5, D III Type 6, D V Type 1, D V Type 2, D V Type 6, D V Type 7, D V Type 8, D VI Type 1, D VI Type 2, D VI Type 3, D VI Type 4, DAR1, DAR2, weak D 4.0, weak D 4.1, weak D 4.2.2, weak D 4.3, weak D Type 29, DBS-0, DBS-1, DBS-2, DFV, DAU5, DBT-1, DBT-2, DOL1, DOL2, DOL3, DCS1, DTO, RHDψ, and other hybrid genes. Exon 7 992A>T and/or 1048G>C DBT1, DBT2, D IVa,D IVa Type 2, D IV Type 3, D IV Type 4, D IV Type 5, D IVb, DBT-1, DBT-2, DHAR, and hybrid genes. Exon G>C and/or 1193A>T D IV Type 3, D IV Type 5, D IVb, DBT2, DHAR, weak D Type 2, weak D Type 41. * Presence of the SNP when there is not a band in the multiplex PCR. Possible D variants when the SNP is detected. Fig. 7. Multiplex PCR analysis on 4% agarose gel. Lanes 1 through 6 correspond to Flow Charts 1 through 6, respectively. into account the influence of ethnicity and the inclusion of control samples of known genotypes. For example, in Austria when all exons are amplified in the multiplex PCR and the sample is C+, we suggest the investigation of weak D Type 1 before weak D Type 3, considering their prevalence. In addition, it would be important to add the Exon 3 sequencing to identify the DFL variant, which is present in this population, and it was not covered by the algorithm herein proposed. The differentiation between partial D and weak D is important for selection of RBC units to conserve stocks of D, and for the administration of RhIG prophylaxis to prevent anti-d formation. Admixed populations, such as the Brazilian population, can present a high quantity of variants. A comprehensive investigation of the RHD alleles that encode weak D expression at the RBC surface could have a considerable impact on the typing and transfusion strategy in countries like Brazil where the prevalence of D phenotype ranges from 5% to 12%, approximately. Therefore, combined with serology, this algorithm using six flow charts based on simple molecular methods can be useful for solving discrepancies and for defining transfusion strategies. This approach is a good solution for services with limited resources and focuses the search onto the population s prevalence, avoiding unnecessary costs that would be incurred if the whole RHD was analyzed. Previous reports have already described other methods and strategies to identify D variants and laboratory-developed tests are still a good option. 4,10,11 Besides that, commercial kits are also available, from kits based on sequence-specific primers PCR that covers a limited numbers of D variants 9 to high-throughput methods, as microarrays. 10,12-14 However, in Brazil, as in other emerging countries, microarray and sequencing are still expensive techniques. Although we could spend more time using the algorithm herein proposed compared with sequencing of all 10 exons of RHD or with the microarray, the cost of this strategy is inferior. We were able to solve 23% of the studied samples with atypical D expression using only one multiplex PCR and one PCR-RFLP, which can be concluded in 8 hours. Additionally, the essential hands-on time to perform a multiplex PCR for one and 10 samples is approximately 30 and 40 minutes, respectively, and to perform a PCR-RFLP is approximately 45 and 55 minutes. Depending on the number and complexity of the samples and the laboratory automation, required time from sample registration until final report may vary from 8 to 72 hours. According to our reality, the use of these two simple protocols (one multiplex PCR and one PCR-RFLP) is approximately 30 times less expensive than sequencing all 10 exons or performing microarray. Additionally, 37% of analyzed samples were identified with one multiplex PCR and two PCR-RFLP, which is approximately 26 times less expensive than sequencing all exons and 13 times less expensive than performing microarray. In total, we could successfully classify 60% of investigated samples without performing any sequencing and without increasing the staff costs. 6 TRANSFUSION Volume **, ** **

7 STRATEGY TO IDENTIFY D VARIANTS TABLE 3. Exons amplified in the multiplex PCR 7 results, excluded and included SNPs, possible variants associated with the results, and flow chart used for final resolution Flow chart Exons amplified in multiplex PCR Excluded SNPs Included SNPs Possible variants 1 Weak D Type 1, Weak D Type 3, Weak D Type 38, Weak D Type 49, DVII, DNB, DIV, DV, DAU0, DAU1, DAU2, DAU3, DAU4, DAU6, DAU7, and DMH. 3, 4, 5, 7, 9 455A>C, 514C>T, 602C>G, 667T>G, 992A>T, 1048G>C, 1154G>C, 1193A>T 1154G>C or/and 1193A>T Weak D Type 2, D IV Type 3, D IV Type 5, and D IVb 2 3, 4, 5, 7 455A>C, 514C>T, 602C>G, 667T>G, 992A>T, 1048G>C 3 3, 7, 9 455A>C, 992A>T, 1048G>C, 1154G>C, 1193A>T 514C>T and/or 602C>G, 667T>G D VI Type 1, D VI Type 2, DAR1, DAR2, weak D 4.0, weak D 4.1, weak D 4.2.2, and weak D T>G RHDψ, DOL1, DOL2, DOL3, DV Type 2, DV Type 1, and DAU5 3, 4, 7, 9 455A>C, 514C>T, 602C>G, 992A>T, 1048G>C, 1154G>C, 1193A>T 7, 9 992A> or/and 1048G>C, 1154G>C or/and 1193A>T 455A>C, 514C>T and/or 602C>G, 667T>G D IIIa,D IIIb,D III Type 6, D VI Type 3, and D VI Type 4 5 4, 5, 9 455A>C, 514C>T, 602C>G, 1154G>C, 1193A>T 667T>G, 992A>T and/or 1048G>C D IVa and D IVa Type 2 6 It is interesting to note that there are some variants identified in this algorithm that are not covered by microarray, as weak D Types 38 and 18, DAU-0, DAU-6, DVII, DMH, and DOL2, which were present in 11.6% of the samples studied. Another benefit includes the range of RhD variants that can be identified. We have identified different types of weak D and partial D in our population that has particularities due to the high degree of miscegenation. Besides the natives who inhabited Brazil before the discovery of Portugal in 1500, the country received immigrants from Portugal, Spain, Germany, Italy, Africa, and other regions of the world. 15 Therefore, this algorithm includes the identification of RHD variants from different ethnic groups. For example, we were able to characterize variants of African origin, like weak D Types 4.0 and 4.2.2, and of European origin, as weak D Types 1, 2, 3, and 38. Additionally, this strategy allowed us to identify variants not previously described in the Brazilian population. DAU0 was found in two samples and it was previously described in Africans from Mali and Congo-Teke, as well as DAU-6 that was identified in six donors and was formerly described in the Congo-Teke population. DAU-5 that we found in one individual was initially observed in a blood donor from Germany, 16 later in the Canadian and Congo-Teke population. 17,18 DNB, a variant from the European cluster found in Germany and Switzerland, 19 was identified in one individual. DOL2 and weak D Type 18, which were identified in two and one sample, respectively, are very rare variants described in only a few individuals. In conclusion, we have designed a simple and practical strategy to investigate samples with unknown RHD status, which is being routinely used in our service. This strategy is useful for resolution of RhD discrepancies and to distinguish weak D from partial D. The algorithm can be adapted to other populations when the admixture of ethnicities and the prevalence of D variants is taken into account. We believe that this strategy can be adopted by services with limited resources interested in the performance of molecular analysis of altered D expression because it is a relatively easy, logical, inexpensive, and efficient strategy. ACKNOWLEDGMENT The authors thank Celso Bianco for his critical review of the manuscript. CONFLICT OF INTEREST The authors report no conflicts of interest or funding sources. REFERENCES 1. Wagner FF, Gassner C, Muller TH, et al. Molecular basis of weak D phenotypes. Blood 1999;93: Volume **, ** ** TRANSFUSION 7

8 ARNONI ET AL. 2. Westhoff CM. The structure and function of the Rh antigen complex. Semin Hematol 2007;44: Flegel WA, Denomme GA, Yazer MH. On the complexity of D antigen typing: a handy decision tree in the age of molecular blood group diagnostics. J Obstet Gynaecol Can 2007;29: Credidio DC, Pellegrino J, Castilho L. Serologic and molecular characterization of D variants in Brazilians: impact for typing and transfusion strategy. Immunohematology 2011;27: International Society of Blood Transfusion (ISBT) Working Party for Red Cell Immunogenetics and Blood Group Terminology. Blood group allele terminology. [cited 2013 Sep]. Available from: red-cell-immunogenetics-and-blood-group-terminology/ blood-group-terminology/blood-group-allele-terminology/ 6. Flegel W. RhesusBase [cited 2013 Sep]. Available from: 7. Maaskant-van Wijk PA, Faas BH, de Ruijter JA, et al. Genotyping of RHD by multiplex polymerase chain reaction analysis of six RHD-specific exons. Transfusion 1998; 38: Legler TJ, Maas JH, Kohler M, et al. RHD sequencing: a new tool for decision making on transfusion therapy and provision of Rh prophylaxis. Transfus Med 2001;11: Cruz BR, Chiba AK, Moritz E, et al. RHD alleles in Brazilian blood donors with weak D or D-negative phenotypes. Transfus Med 2012;22: Brajovich ME, Boggione CT, Biondi CS, et al. Comprehensive analysis of RHD alleles in Argentineans with variant D phenotypes. Transfusion 2012;52: Fichou Y, Le Marechal C, Jamet D, et al. Establishment of a medium-throughput approach for the genotyping of RHD variants and report of nine novel rare alleles. Transfusion 2013;53: Wang D, Lane C, Quillen K. Prevalence of RhD variants, confirmed by molecular genotyping, in a multiethnic prenatal population. Am J Clin Pathol 2010;134: Granier T, Beley S, Chiaroni J, et al. A comprehensive survey of both RHD and RHCE allele frequencies in sub- Saharan Africa. Transfusion 2013;53: Kappler-Gratias S, Auxerre C, Dubeaux I, et al. Systematic RH genotyping and variant identification in French donors of African origin. Blood Transfus 2013;17: Curtin P. Atlantic slave trade: a census. Milwaukee (WI): University of Wisconsin Press; Chen Q, Flegel WA. Random survey for RHD alleles among D+ European persons. Transfusion 2005;45: Denomme GA, Wagner FF, Fernandes BJ, et al. Partial D, weak D types, and novel RHD alleles among 33,864 multiethnic patients: implications for anti-d alloimmunization and prevention. Transfusion 2005;45: Touinssi M, Chapel-Fernandes S, Granier T, et al. Molecular analysis of inactive and active RHD alleles in native Congolese cohorts. Transfusion 2009;49: Wagner FF, Eicher NI, Jorgensen JR, et al. DNB: a partial D with anti-d frequent in Central Europe. Blood 2002;100: TRANSFUSION Volume **, ** **