Journal Logo

Original Articles: Immunobiology and Genomics

Prediction of Crossmatch Outcome of Highly Sensitized Patients by Single and/or Multiple Antigen Bead Luminex Assay

Vaidya, Smita1,4; Partlow, David1; Susskind, Brian2; Noor, Maryam1; Barnes, Titus1; Gugliuzza, Kristine3

Author Information
doi: 10.1097/


End-stage renal disease (ESRD) patients sensitized by alloimmunization such as previous pregnancies, multiple transfusions, and/or allograft rejection often develop anti-human leukocyte antigen (HLA) antibodies against a large percentage of HLA antigens as evidenced by monthly HLA antibody screen (1–5). Currently, there is no particular strategy advocated by the United Network for Organ Sharing (UNOS) to find crossmatch-compatible donors for highly sensitized patients, except that UNOS encourages transplant centers to define unacceptable antigens when listing their sensitized patients for deceased donor transplantation. However, definition of unaccepted antigens on the basis of HLA antibody specificities has been historically limited due to lack of appropriate reagents and technologies. Consequently, in many centers these patients are crossmatched with either all available ABO-compatible donors or they are crossmatched only with well-matched donors. Both strategies are costly and inefficient: the first results in performing many unnecessary crossmatches with no net increase in finding crossmatch-negative donors; the second reduces the probability of finding crossmatch-negative donors owing to the highly polymorphic nature of the HLA system.

In this retrospective analysis, we used a Luminex-based assay to identify HLA Class I and II specificities in the sera of 20 highly sensitized patients and applied this information to predict T and B cell flow cytometry crossmatch outcomes with greater than 90% accuracy.


Selection of Patient Sera

Twenty highly sensitized patients immunized by prior allograft rejections, multiple pregnancies, and/or multiple transfusions were selected for this study. Sera collected from them for screening against deceased donors had greater than 70% HLA Class I and/or II panel reactive antibodies (PRA) and PRA values did not fluctuate appreciably over a period of at least two years (±10%). Similarly, anti-HLA antibody specificities remained consistent during the same time frame.

HLA Antibody Screen by Luminex

Serum samples collected from these 20 patients were analyzed for HLA class II and I PRA and specificities by Tepnel Lifecode Luminex system (Stamford, CT). In brief, HLA Class I or II antigen-coated Luminex beads were incubated with patient’s serum sample in 96-well Millipore multiscreen filter plates (Bedford, MA). Unbound excess serum was removed by washing the beads in the wells after which phycoerythrin-conjugated goat antihuman immunoglobulin (Ig) G antibody was added. The plates were incubated in the dark on a rotating platform. Following incubation, the test samples were analyzed on Luminex instrument. All assays included both in-house as well as manufacturer’s positive and negative controls.

To determine if an individual bead was positive, the median fluorescent intensity (MFI) of the bead was divided by the MFI of three negative control beads and a background adjustment factor subtracted to provide three different adjusted ratios. A mathematical negative control calculation gave a fourth negative ratio. A specific bead was considered positive if two or more adjusted ratios were positive.

Serum analysis was done by the chi-square method. The tail was sorted by chi square, R value, and MFI. Because antibodies to the high-frequency HLA Class I and II epitopes are predominant in sensitized patients, the first set of analysis was done with the expectation that antibody specificities would be restricted to one or two of the known public determinants (6–9, 12–19). Table 1 shows the major crossreactive patterns observed among the HLA-A, -B, -DR, -DRw, and -DQ gene products. Score values were calculated for each of the specificities as the percentage of total reactivities.

Nomenclature used in assigning HLA specificities

The sera with the score values of less than 75% were further tested using single antigen beads (Lab Screen Single Antigen Class I and II Antibody Detection System, One Lambda, Canoga Park, CA).

To determine if an individual bead was positive, the MFI of each HLA specific bead was normalized with the MFI of the negative control beads and adjustment was made for the background MFI caused by negative control serum using the mathematical formula provided by the manufacturer. This formula allows converting the strength of reaction to the familiar serologic scoring range of 8, 4, 2, and 1, where 8 is a strong positive reaction, 4 is weak positive, 2 is considered doubtful positive, and 1 is a negative reaction. For our analysis, beads showing only 8 and 4 reactions were taken into consideration. The single-antigen bead analysis identified private as well as public specificities in the sera and score values increased from 75% to 90%.

Crossmatch Protocol

Information obtained from Luminex serum analysis was applied to predict deceased donor crossmatches (XM) retrospectively. Two types of crossmatches were analyzed. All of these sera were crossmatched with blood group compatible deceased donors’ T and B lymphocytes in a complement-dependent cytotoxicity (CDC) crossmatch. The CDC T cell XM was enhanced with antihuman immunoglobulin (AHG). The source of lymphocytes for the CDC XM was preharvest blood, drawn prior to organ recovery, 90% of the time. Sera were further tested in flow T- and B-cell XM (FLXM) using the pronase technique of Vaidya et al. (10). Source of the lymphocytes for FLXM was either lymph nodes or spleen. Overall, 2,331 T-cell CDC, 2,198 B-cells CDC, 236 T-cell FLXM, and 249 B-cell FLXM were analyzed using 239 deceased donors for the CDCXM and 38 donors for FLXM. Each serum was tested undiluted and in duplicate in both CDC and FLXM.

Statistical and Data Analysis

Fisher exact test was used for statistical analysis. P value ≤0.05 was considered significant. Observed and predicted XM data were divided into four categories: observed (O) and predicted (P) positive (O+/P+), observed negative but predicted positive (O−/P+), observed positive but predicted negative (O+/P−), and observed and predicted negative (O−/P−). The predictive values of XM were calculated as follows:

Sensitivity = (O+/P+)/[(O+/P+)+(O−/P+)]

Specificity = (O−/P−)/[(O−/P−)+(O+/P−)]

Efficiency = (O+/P+)+(O−/P−)/N


PRA and Antibody Specificities

Table 2 shows the PRA values and anti-HLA antibody specificities identified by Luminex system for the 20 patients whose sera were selected for this study. HLA specificities were identified by multiantigen beads in 16 sera. Specificities in remaining four sera (nos. 1, 9, 12, and 20) that could not be clarified by multiantigen beads alone were further tested by single antigens beads. The majority of the sera had reactivities against high frequency Class I and Class II public epitopes. However, nine of the sera (nos. 1, 6, 10, 11, 13, 14, 16, 19, and 20) also showed additional reactivities against private epitopes. The score values shown in Table 2 are the percentages of total reactivities accounted for by given specificity assignment. The average scores for the Class I and II specificity assignments were 93% and 89%, respectively. Low score values indicate the presence of additional antibody specificities yet to be identified. The sera from the 20 patients analyzed in this project had strong anti-HLA titer with endpoint dilution on average greater than 1:64 and MFI of undiluted sera was equal to or greater than 18,000±2,500.

Serum analysis

Observed and Predicted XM Comparisons

CDC Crossmatches

A total of 2,331 T-cell and 2,198 B-cell CDC XMs were retrospectively analyzed for this study. The predictive values for these crossmatch results were determined by comparing the antibody specificities identified by the Luminex serum analysis with the HLA phenotype of the deceased donor. Table 3 shows the comparison between observed T and B cells XM results and the predicted outcomes by Luminex serum analysis. Of the total 2,331 T cell AHG XM, Luminex serum analysis correctly predicted 1,426 positive and 378 negative XM. However, 283 positive XM were incorrectly predicted to be negative XM by Luminex serum analysis, resulting in an overall efficiency of Luminex analysis of 77% for prediction of the CDC T-cell XM results. These results can be broken down further as follows: Sensitivity of the Luminex analysis for predicting positive T-cell CDC XM was 85%. Specificity of the Luminex analysis (i.e., predicting negative crossmatch results because no donor-specific anti-HLA antibody was identified) was 57% (possibly due to the presence of non-HLA antibodies).

Comparison of observed crossmatch outcomes with predicted outcomes by Luminex serum analysis

Similarly, Luminex serum analysis correctly predicted 1,649 B-cell CDC (1,486 positive and 163 negative) XM results with 75% efficiency. Sensitivity of Luminex analysis in accurately predicting positive B-cell CDC XM was just as high as T-cell CDC XM (85% versus 87%); however, the specificity of Luminex analysis in predicting negative B-cell XM was only 33% indicating that the B-cell CDC XM was positive more often in absence of identified anti-HLA Class II antibodies. The lower specificity may come from the fact that B cell CDC XM can be positive due to donor-specific HLA Class I and Class II antibodies (11), whereas our analysis was based only on the anti Class II as detected by Luminex.

T- and B-cell Flow Crossmatches

A total of 236 T- and 249 B-cell FLXM were analyzed retrospectively for this study (Table 3). The Luminex serum analysis accurately predicted 96 positive and 129 negative T cell FLXM resulting in 95% efficiency. Similarly, Luminex serum analysis accurately predicted 123 positive and 113 negative B-cell FLXM resulting in 95% efficiency. The difference in efficiency between CDC and FLXM was statistically significant (P=0.02 for T-cell XM and P=0.01 for B-cell XM). The sensitivity of Luminex for prediction of positive T- and B-cell FLXM was 98% each. The specificity of Luminex analysis in predicting negative T and B cell FLXM was 93% and 91%, respectively. Interestingly, the difference in sensitivity values between CDC and FLXM was not statistically significant (P=0.4 for T- and B-cell XM), whereas the difference in specificity values between CDC and FLXM was highly significant (P=0.0004 for T-cell XM and P=1.6×10−10 for B-cell XM).


The main objective of this study was to determine if Luminex serum analysis can accurately predict deceased donor T- and B-cell CDC and flow crossmatches for highly sensitized ESRD patients. This objective can be achieved only if HLA antibody specificities in the sera can be correctly identified so that crossmatch results correlate with the phenotypes of the donors. Identification of HLA antibody specificities in highly sensitized patients can be difficult because these sera appear to contain antibodies to many HLA antigens. However, several studies (6–9, 12–19) have shown that highly sensitized patients often develop HLA antibodies to the high-frequency public HLA determinants. In this study sera of highly sensitize patients were analyzed by Luminex assay using micro particle beads coated with single and/or multiple HLA antigens. Two types of crossmatches were analyzed: T- and B-cell flow and CDC XM. The data showed that Luminex serum analysis done by single and/or multiantigen beads can predict T and B cell deceased donor FLXM with 95% efficiency. The analysis resulted in extremely low levels of O+/P− (4% for each T cell and B cell) and O−/P+ (1% for each T cell and B cell) FLXM. In contrast to FLXM, Luminex serum analysis was far less accurate in predicting CDC crossmatches. The false positive rate of T- and B-cell CDC XM was 12% and 15% and false-negative rates were 10% each.

It is noteworthy that O+/P− CDC crossmatch rates were high in light of the fact that anti HLA specificities were identified by a highly sensitive technique and therefore one should expect a low frequency of O+/P− results. Possible explanation may be as follows. First, as indicated in the Methods Section, CDC crossmatches were performed using lymphocytes isolated from preharvest blood of deceased donors. These lymphocytes are more susceptible to complement-mediated cytotoxicity than lymphocytes isolated from normal blood donors and thus nonspecific killing of lymphocytes may contribute to the O+/P− rate. (The amount of time that donors spend on respirator before the blood is harvested appears to contribute to the likelihood of O+/P results [S Vaidya, unpublished observations].) Second, the CDC screening crossmatches were done without IgM reducing agents such as dithiothreitol (DTT), except for those patients with lupus. It is possible that many nonlupus patients may have developed auto antibodies over time and therefore one should perform the CDC XM with DTT on a routine basis to avoid non-HLA IgM antibodies. Third, other complement-fixing non-HLA antibodies may also have contributed to the O+/P− rate, especially because these were highly allosensitized patients to begin with. Finally, HLA specificities may have been missed by Luminex, which were responsible for making CDCXM positive, especially IgM antibodies. However, when we evaluated these sera screened in the past by CDC method using fresh panel cells no additional HLA specificities were found. Therefore, the first three reasons above are most likely explanations.

However, in the light of above explanations, highly sensitized patients should not be transplanted on the basis of negative CDC XM alone, particularly if the source of lymphocytes is preharvest blood, because likelihood of false-positive results may be high.

One of the most important reasons for doing serum analysis is its accuracy in predicting negative crossmatches. These data show that Luminex serum analysis is far more accurate in predicting negative T and B cell FLXM than CDC XM, as evidenced by its specificities (93% and 91% versus 57% and 33%, respectively). This is a very useful information in clinical setting because if donors are selected by actually performing crossmatches between every donor with every (blood group compatible) recipient, the frequency of finding negative T and B FLXM by this process of random selection would be 56% ([FN+TN]/n=131/236) and 46% (115/249; Table 3), respectively. In contrast, frequencies of negative T- and B-cell FLXM increased to 93% (TN/ [TN+FP]=129/138) and 91% (113/124; Table 3), respectively, when Luminex serum analysis was used to preselect the donors. Furthermore, preselection process reduces the number of T-cell FLXM from 236 to 138 (58%) and B-cell FLXM from 249 to 124 (50%).

In most centers, deceased donor crossmatch protocol is generally a two-step process. The first step is CDC T-cell screening XM (AHG) of ABO-compatible recipients to eliminate patients who are clearly crossmatch positive. Sera of the remaining screening crossmatch-negative patients go through a second round of crossmatches if the patients are medically suitable for transplant and especially if they are highly sensitized. The second round of crossmatch generally includes FLXM. Analysis of HLA antibodies in the sera of highly sensitized patients has significant clinical utility in that this analysis can be used to prospectively select donor- recipient pairs without performing unnecessary crossmatches. In fact, this analysis can eliminate the need for a two-step protocol for the selection of deceased donor- recipient pairs. If FLXM is the final crossmatch needed to prove crossmatch compatibility, then there is no need to perform CDC screening XM at all.

Patients selected for this study had maintained their PRA levels over a period of two years with fluctuations of no more than 10%. Similarly, their antibody specificities also remained stable. These data support earlier studies (8, 9, 12) in which authors found that in highly immunized patients monthly antibody specificity analysis is not necessary as long as PRA levels remain stable. On the basis of their data and ours, we recommend that detailed antibody specificity analysis should be carried out only when PRA levels of sera change more than 10%. However, it is important to collect and measure PRA of the sera of highly sensitized patients on a monthly basis to ensure the stability of PRA because sensitizing events such as blood transfusions, infections, and/or transplant nephrectomies are not communicated to the tissue-typing labs on a regular basis.

A computer-based approach called HLAMatchmaker for determining HLA compatibility between kidney donor-recipient on the basis of amino acid sequence differences has been proposed by Duquesnoy et al. (20, 21). However, reports of Claas et al. (22) and Lauxet et al. (23) have shown that its utility is limited to those patients who have common HLA phenotypes. HLA polymorphism does not allow patients with uncommon HLA phenotypes to profit from such approaches.

Luminex analysis is very sensitive in that it can detect antibodies that are nonreactive in FLXM. At this point, our highly sensitized patients are transplanted when donor- specific FLXM is negative even though extremely low-titer antibody is present as detected by Luminex. We are currently evaluating the effect of Luminex+/FLXM– antibodies on graft survival.

This serum analysis is also a valuable tool in sharing kidneys over a wider geographic region. One of the major obstacles in sharing kidneys over wider geographical region is the increase in cold storage time of the harvested kidneys, while necessary crossmatches are being done by the tissue-typing labs. In addition to an increase in ischemia time, shipping of the tissues for the crossmatch purpose is also costly. Therefore an analysis method that can predict the final FLXM outcome with 95% accuracy would be extremely useful and would significantly increase the chances of transplanting imported kidneys to the designated highly sensitized recipient candidate. Moreover, highly sensitized patients may not have to wait for zero-mismatched kidneys in order to be transplanted.


1. Vaidya S. Synthesis of new and memory HLA antibodies from Acute and chronic rejections versus pregnancies and blood transfusions. Transplant Proc 2005; 37: 648.
2. Gebel HM, Bray RA, Nickerson P. Pre-transplant assessment of donor-reactive, HLA specific antibodies in renal transplantation: Contraindication vs. risk. Am J Transplant 2003; 3: 1488.
3. Cai J, Terasaki P. Humoral theory of transplantation: Mechanism, prevention and treatment. Hum Immunol 2005; 66: 334.
4. Vongwiwatana A, Tasanarong A, Hidalgo LG, Halloran PF. The role of B cells and alloantibodies in host response to human organ allografts. Immunol Rev 2003; 196: 197.
5. Cecka JM, Zhang Q, Reed EF. Preformed cytotoxic antibodies in potential allograft recipients: recent data. Hum Immunol 2005; 66: 343.
6. Oldfather JW, Anderson CB, Phelan DL, et al. Prediction of crossmatch outcome in highly sensitized dialysis patients based on the identification of serum HLA antibodies. Transplantation 1986; 42: 267.
7. Oldfather JW, Mora A, Phelan DL, et al. The occurrence of cross reactive “public” antibodies in the sera of highly sensitized dialysis patients. Transplant Proc 1983; 15: 1212.
8. Schwartz BD, Luehrman LK, Lee J, Rodey GE. A public antigen determinant in the HLA B5 cross-reactive group. A basis for cross-reactivity and a possible link with Behecet’s disease. Hum Immunol 1980; 1: 938.
9. Schwartz BD, Luehrman LK, Lee J, Rodey GE. Public antigenic determinant on a family of HLA-B molecules. Basis for cross-reactivity and a possible link with disease predisposition. J Clin Invest 1979; 64: 938.
10. Vaidya S, Cooper TY, Avandesalehi J, Barnes T, et al. Improved flow cytometric detection of HLA alloantibodies using pronase. Transplantation 2001; 71: 422.
11. Pellegrino MA, Belvederes M, Pellegrino AG, et al. B cell peripheral blood lymphocytes express more HLA class I antigens than T cells peripheral blood lymphocytes. Transplantation 1978; 25: 93.
12. Middleton D, Williams F, Meenagh A, et al. Analysis of distribution of HLA-A alleles in populations from five continents. Hum Immunol 2000; 61: 1048.
13. Ayres J, Cresswell P. HLA-B specificities and w4, w6 specificities are on the same polypeptide. Eur J Immunol 1976; 6: 794.
14. Collins MM, Tang T, Slack R, et al. The relative frequencies of HLA- DRB1*01 alleles in the major US populations. Tissue Antigen 2000; 55: 48.
15. Baldassarre LA, Steiner NK, Jones P, et al. Limited diversity of HLA-DRB1*02 alleles and DRB1-DRB5 haplotype associations in four United States population group. Tissue Antigen 2003; 61: 249.
16. Tang TF, Wang J, Slack R, et al. DRB1*03 diversity and DRB3 associations in five major population groups in the United States. Hum Immunol 2002; 63: 221.
17. Chen DS, Ting F, Tang T, et al. Relative HLA DRB1*04 allele frequencies in five United States populations found in a hematopoietic stem cell volunteer donor registry and seven new BDB1*04 alleles. Hum Immunol 2002; 63: 665.
18. Gans CP, Tang TF, Slack R, Ng J, et al. DRB1*14 diversity and DRB3 associations in four major population groups in the United States. Tissue Antigen 2002; 59: 364.
19. Rizzuto G, Li L, Steiner N Slack R. et al. Diversity within the DRB1*8 allele family in four populations from a United States hematopoietic stem cell donor database and characterization of five novel DBR1*8 alleles. Hum Immunol 2003; 64: 607.
20. Duquesnoy RJ. HLAMatchmaker; a moleculary based donor selection algorithm for highly alloimmunized patients. Transplant Proc 2001; 33: 493.
21. Duquesnoy RJ. HLAMatchmaker; a moleculary based algorithm for histocompatibility determination. I. Description of algorithm. Hum Immunol 2002; 63: 339.
22. Claas FH, Witvliet MD, Duquesnoy RJ, et al. The acceptable mismatch program as a fast tool for highly sensitized patients awaiting a cadaver kidney transplantation: short waiting time and excellent graft outcome. Transplantation 2004; 78: 190.
23. Laux G, Mytilineos J, Oplelz G. Critical evaluation of the aminoacid triplet-epitope matching concept in cadaver kidney transplantation. Transplantation 2004; 77: 902.
24. Tang TF, Huang AY, Pappas A, et al. Relative frequencies of DRB1*11 alleles and their DRB3 associations in five major population groups in a United states bone marrow registry. Hum Immunol 2000; 61: 820.
25. Sintasath DM Tang TF, Slack R, et al. Relative HLA-DRB1*13 allele frequencies and DRB3 associations of unrelated individuals from five US populations. Hum Immunol 1999; 60: 1001.

Complement dependent cytotoxicity crossmatch; Flow cytometry crossmatch

© 2006 Lippincott Williams & Wilkins, Inc.