The Italian Quality Control Scheme for Crossmatching Procedures and HLA Sera Screening: The 2002 Pilot Study

The Turin Transplant Immunology Service is the reference center for transplantation in the northwestern Italian region of Piedmont. Since 1991, this service has organized a regional quality control (QC) program that supervises the laboratories that perform HLA typing of bone marrow donors in Piedmont. The aims of this program were, first, to ensure the high quality of the HLA typing laboratories and, second, to provide frequent technical updates. Initially, the Piedmont region established its own parameters and criteria for performing regional QC, but in 1995 the European Federation for Immunogenetics (EFI) regulations were adopted1 and QC of HLA typing became nationwide in 1998.

The 2000 update of the EFI standards strongly recommended extending proficiency testing to both crossmatch and antibody reactivity,2 and successful participation in external QC programs became mandatory for EFI accreditation. Therefore, the Italian National Transplantation Centre decided to implement a nationwide QC plan in 2002, which the Turin Transplant Immunology Service was asked to both organize and participate in.

Seventeen laboratories located throughout Italy took part in this QC exercise. All met national guidelines for laboratory performance, and all were part of the national Bone Marrow Donor Registry. Laboratories were required to perform serologic analyses by crossmatching and anti-HLA antibody determination. Complement-dependent cytotoxicity (CDC) remains the classic method of crossmatching evaluation and was used by all participating laboratories, although some have adopted more sensitive methods based on flow cytometry (for crossmatching) and enzyme-linked immunoassay (ELISA, only for antibody detection). However, since the number of national laboratories that use these techniques is as yet too small for significant comparisons, the present study will describe only the results obtained by CDC.

The study was designed to satisfy EFI standards, since it was part of the biennial accreditation process. These standards require laboratories to screen 12 serum samples and to perform at least 30 crossmatches a year. The study reported herein was based on the analysis of 6 anti-HLA antibody–negative serum samples and 6 positive serum samples that contained monospecific or polyspecific anti-HLA antibodies. As per EFI standards, to be successful, a laboratory could report 4 or fewer discrepancies in the crossmatch assays and 2 or fewer discrepancies in antibody determination.3 

The QC analysis used 12 serum specimens (numbered 1–12) and 6 blood samples (labeled A–F), which were distributed by the organizers to all laboratories in 2 shipments of 6 serum samples and 3 blood samples. Sample size was 1 mL per serum sample and 15 mL per blood sample. The serum samples (6 were anti-HLA antibody negative and 6 were positive with monospecific or polyspecific anti-HLA antibodies) and blood specimens were characterized (Tables 1 and 2) and chosen by the organizers to provide both positive and negative samples for crossmatch analysis. Blood was obtained from volunteer donors and collected in ACD Vacutainer tubes (Becton Dickinson, Plymouth, UK).

The results of crossmatch analyses were scored as follows: (1) a positive crossmatch result could be obtained using either total lymphocytes or purified T-cell and/or B-cell populations; (2) the consensus was defined as a result in agreement with at least 75% of the participating laboratories; and (3) a discrepancy occurred when a crossmatch result differed from the 75% consensus. Crossmatch analysis obtained after dithiothreitol treatment of sera was routinely performed in only 8 of 17 laboratories; consequently, the relative results are not included herein.

Of the 17 participating laboratories, 13 performed screening for anti-HLA antibodies by CDC. The other 4 laboratories performed this procedure using flow cytometry or ELISA, so their results are not included herein (Table 3). Negative serum specimens were defined as those with a panel reactive antibody (PRA) of less than 10% and no detectable antibodies. The following results were considered discrepant: (1) when a result contrasted with the 75% consensus; (2) when a reported antibody specificity differed from both the 75% consensus and the specificity established by the organizer; and (3) no antibody specificity of the sample was reported (indicating polyspecificity), in contrast with the 75% consensus and the organizer's results.

All laboratories were required to perform 18 CDC crossmatches but differed in their use of total or fractionated lymphocytes (Table 4). The target of 75% consensus among participating laboratories was reached in 32 of 36 crossmatches. In detail, of the 17 laboratories, 9 (52%) made no errors, 4 (24%) made 1 error, and 4 (24%) made 2 errors. Thus, all laboratories satisfied the standards of the EFI and, overall, crossmatch analysis showed good interlaboratory reproducibility. The negative note was that not every laboratory performed all 36 crossmatches because blood samples had deteriorated because of inappropriate storage.

Four crossmatches (A/2, C/2, C/3, and E/7) did not reach the consensus (Table 5). Most of the laboratories that performed crossmatches with total lymphocytes scored A/2, C/2, and C/3 as negative, whereas laboratories that separated T and B lymphocytes obtained a mostly positive B-cell crossmatch result. The explanation for these discrepant results lies in the presence of anti-HLA class II antibodies, which were subsequently detected in some of the laboratories by means of CDC, ELISA, or flow cytometry (note that the sera had not been analyzed for the presence of anti-HLA class II antibodies by the organizers before shipping). It is known that B lymphocytes represent approximately 20% of the total lymphocyte population; therefore, in the presence of anti-HLA class II antibody, crossmatch with separated B cells should be more suitable to detect such antibodies, since the B-cell population is more enriched.

A positive crossmatch result for E/7 was mostly obtained using B lymphocytes alone. The antibody present in serum 7 was anti-HLA-B5 and sample E was HLA-B5 positive; therefore, the result of the crossmatch with total lymphocytes should have been positive. Since 60% of participants found a negative reactivity with total lymphocytes, this anti-HLA-B5 antibody is supposed to be weakly reactive. Some authors have shown that a B-cell–positive crossmatch is not only due to anti-HLA class II antibodies but also due to weak or low-titer anti-HLA class I antibodies.4,5 In fact, weak class I antibody reactivity can be detected better on B lymphocytes, since they present a higher HLA antigen density than T cells.6 These considerations should explain the lack of consensus in the A/2, C/2, C/3, and E/7 crossmatches, and they underline the importance of performing crossmatches with separated T and B cells to obtain the maximum amount of information about the serum under scrutiny.

The D/7 crossmatch has reached consensus, but contrary to our expectations, it was scored as negative instead of positive. In fact, the serum presented HLA antibodies against B5 and B35, whereas the cells were typed as HLA-A11, A30, B18, and B35. Since this result was obtained by all participating laboratories, we deduce that loss of antibody reactivity of the serum occurred.

The results of serum antibody detection of individual laboratories are given in Table 6. The table also shows that some laboratories identified additional specificities besides the principal one assigned by the organizers. In such cases, the result was not considered discrepant. Among the 13 laboratories that regularly performed antibody analysis by CDC, 7 (54%) satisfied the EFI standards. In detail, of the 13 laboratories, 4 (31%) made no errors, 2 (15%) made 1 error, 1 (8%) made 2 errors, and 6 (46%) made more than 2 errors. In summary (Table 7), 32% of the errors were due to incorrect assignments, whereas 68% were due to missed antigens.

Quality control of HLA typing laboratories is required by organizations such as the EFI and the American Society for Histocompatibility and Immunogenetics. This article reports on the first QC exercise of HLA typing laboratories in Italy, which evaluated CDC crossmatching and serum antibody determination in 17 laboratories. The importance of proper execution of the crossmatch test is clear: a negative result will favor transplantation, whereas a positive result will give an indication of the risk of hyperacute graft rejection. Crossmatch testing has several, potentially dangerous, pitfalls, such as false-negative and false-positive results.

A false-negative crossmatch result may lead to acute rejection of the transplant,7–9 and technical errors or defective batches of complement may occur because of low lymphocyte reactivity with the test sera. Donor lymphocytes are usually obtained from individuals in critical condition, and lymphocyte reactivity may be compromised,10 making result interpretation difficult. To solve this problem, our experience suggests performing CDC crossmatching with lymphocytes isolated from donor lymph nodes rather than from the spleen or peripheral blood, since a great proportion of nodal lymphocytes are in the early stages of their life cycle and less exposed than blood or spleen lymphocytes to biological stress.11 

Similarly, false-positive crossmatch results must also be avoided, since they can prevent a patient from undergoing transplantation; therefore, an explanation must be found for every positive crossmatch result. Thus, it is important (1) to determine the antibody specificity of the serum to uncover possible specific anti-HLA immunization of the patient; (2) to know if recipients have autoantibodies that are responsible for the positive crossmatch result, since these are not detrimental for the transplant; and (3) to evaluate the PRA, which indicates the degree of immunization of a patient and can help to choose the most adequate posttransplantation therapy or even the donor assignment. Indeed, better graft survival has been reported in patients with a PRA of less than 50% who received an HLA-mismatched transplant.12 Another study based on a large cohort of patients reported similar results when the high sensitization was mostly due to both anti-HLA class I and II antibodies.13 For all these reasons, the need for QC of the typing procedure is evident.

The crossmatch results observed in the present study are similar to those reported by Marrari and Duquesnoy.14 In their 12-year QC study, these authors reported a greater than 90% consensus among participants and 71% consensus for the 120 crossmatches performed by 143 laboratories. In comparison, our report reveals far more discrepancies in antibody specificity assignment, with only 54% of participating laboratories satisfying the EFI standards. Several hypotheses can be advanced to explain this result. First, variability in complement cytotoxic activity is a known variable,15 as has been reported in other QC surveys.14,16,17 Second, alterations in serum reactivity caused by improper handling after arrival in the laboratory should not be underestimated. Third, not only the size but also the HLA antigen representation of the cell panel used for PRA screening should be taken into great account. Some laboratories that incurred errors used wide panels (>40 cells); therefore, we presume that some HLA specificities were underrepresented. On the other hand, panels composed of smaller cell numbers may offer a better variety of HLA antigens, explaining how laboratories with panels of 30 cells or fewer made 0 to 1 error. In our experience, however, the number of errors does not correlate with panel size, as shown in the Figure, where the composition of the panel used for PRA is compared with the number of discrepancies per laboratory. In the study by Marrari and Duquesnoy,14 the panel size ranged from 15 to 100 cells, and the average consensus for antibody specificity assignment was satisfactory. Therefore, we confirm that panel size alone does not determine good results; our experience suggests using approximately 40 cells and guaranteeing a sufficiently broad representation of HLA antigens.

We also found that the process of interpreting the results can be complex. It may be difficult to correlate serum reactivity with antigen specificity or the correlation may differ, depending on whether computer-based or manual interpretation is adopted. For this reason, the interpretation of results requires much care and experience.

On several occasions, a serum sample was classified as polyspecific or its specificity could not be defined. In these circumstances, the samples must be diluted, especially when dealing with highly allosensitized patients with a starting PRA of more than 80%. Dilution may cause loss of some antibody specificities, but it helps define the most important ones, which would otherwise escape definition in a polyspecific and highly reactive serum.

The consensus reached for the negative serum samples was 100%. Among the 6 positive serum samples that contained anti-HLA antibodies, 5 were polyspecific and 1 was specific for HLA-A2, the latter being the only one that reached 75% consensus in antibody specificity determination. Similar results were obtained by Duquesnoy and Marrari,16 whose study produced a good consensus rate only with monospecific sera.

The PRA data provided by each participant were not evaluated, because they were strongly influenced by the variables mentioned herein, such as complement strength and cell panel composition. This variability undoubtedly represents one of the unsolvable disadvantages of the CDC technique, which cannot be standardized and consequently suffers from such poor reproducibility that Doxiadis et al18 even suggested reconsidering the importance of the PRA in patient classification.

Discordant PRA results have also been observed in other studies. The 7-year analysis performed by Duquesnoy and Marrari16 showed a relevant variability among the PRA values, whereas in a previous QC study by Howden et al17 the participating laboratories were in concordance with one another. The report by Howden et al is positive but is based on the examination of 6 serum samples in 6 laboratories, whereas the EFI standards recommend a minimum number of 15 participants to reach significance. Therefore, the PRA results should always be taken into account by the transplant centers to make PRA results more comparable in future QC studies. We conclude that there is room for improvement in our national typing procedures and greater cooperation among participating laboratories is necessary to reach satisfactory results that translate into better patient care.

This study was in part supported by Piedmont Region grant (year 2002, No. 396, project entitled New Donor and Recipients Evaluation Procedures to Improve Transplant Quality) and Health Ministry grant (year 2002, No. 3AM/F11, project entitled Organ Transplantation in Elderly People: Epidemiology and Clinical Results). We acknowledge the participating laboratories and their heads who provided the results: Prof Domenico Adorno, MD, CNR Institute, L'Aquila; Biagio Favoino, biologist, Policlinico Hospital, Bari; Roberto Conte, MD, Sant'Orsola Hospital, Bologna; Prof Carlo Carcassi, MD, Binagli Hospital, Cagliari; Vittorio Fossombroni, MD, Firenze-Careggi Hospital; Angelo Nocera, MD, San Martino Hospital, Genova; Enzo Santospirito, biologist, Regional Referring Transplant Centre Matera; Francesca Poli, biologist, Major Hospital, Milano; Marianna Resse, biologist, University Policlinico II, Napoli; Prof Francesco Dieli, MD, Regional Referring Transplant Centre, Palermo; Prof Mario Savi, MD, Medical Clinic Department, Parma; Prof Cesare Gambelunghe, MD, Regional Referring Transplant Centre, Perugia; Maria Luciana Mariotti, biologist, Cisanello Hospital, Pisa; Valerio Misefari, biologist, Riuniti Hospitals, Reggio Calabria; Antonina Piazza, MD, Tor Vergata University, Roma; and Mirella Mariani, MD, National Centre for Blood Transfusion C.R.I., Roma.