A major challenge that most renal transplant centers face today is transplantation of patients with high levels of human leukocyte antigen (HLA) antibodies. These sensitized patients are less likely to be compatible with the random donor pool when compared with nonsensitized patients and not surprisingly, the wait time to transplant for these patients is significantly longer. The concept of adapting strategies to transplant sensitized patients is hardly novel as defining “acceptable mismatches” and hyperimmune trays have been used by Eurotransplant for more than 20 years with high levels of success (1–3). Variations of these strategies have been adapted in the United States, which has resulted in defining “unacceptable antigen” profiles and “virtual crossmatching” for this cohort of patients (4–7). Virtual crossmatching is defined as the ability to predict the outcome of the final crossmatch based on HLA antibody screening and identification. Predictability of the final crossmatch has been problematic in the past because of less specific and sensitive techniques to clearly identify HLA class I and II antibodies. Recent technological advances in HLA antibody identification using solid-phase multiplexing platforms have overcome many of the sensitivity and specificity issues allowing for more accurate prediction (8, 9). Furthermore, utilization of single antigen (SA) beads has provided clear identification of HLA A, B, Cw, DR, DQ, DRw, and DP antibodies which was previously hindered by masking because of cell panel linkage disequilibrium.
The virtual crossmatch protocol was implemented on October 2006 at our transplant center for all sensitized renal patients waiting for deceased donor transplantation. The rationale for moving to virtual crossmatching was dual-fold: (1) use methods that would mimic the sensitivity of the final crossmatch and (2) reduce the time and cost for deceased donor work-up by no longer performing preliminary crossmatch trays. By removing the preliminary crossmatch trays from the deceased donor work-up, the average turnaround time was reduced by 3 hr allowing for quicker organ allocation with reasonable assurance that the sensitized patients at the top of the points list will have a negative final crossmatch.
The objective of this study was to determine if the virtual crossmatch protocol increased access to transplantation for the sensitized patient and to determine the cost savings of the protocol change. This article will summarize the findings of our patient data reviewed from October 2006 to June 2008 in comparison with the Scientific Registry of Transplant Recipients (SRTR) data for 2006.
MATERIALS AND METHODS
This study is a retrospective chart and database review performed with Institutional Review Board approval. All kidney transplant recipients undergoing transplantation from deceased donors between October 1, 2006 and June 30, 2008 were included.
HLA typing antibody screening was performed using FlowPRA screen beads (OneLambda) for the presence or absence of HLA class I and II antibodies. HLA antibody specificity was determined on the Luminex platform using SA bead technology. SA beads for HLA class I included A, B, and Cw loci and class II SA beads included DR, DQ, DRw, and DP loci. All final crossmatches were performed by flow cytometry. The panel-reactive antibody (PRA) levels presented in this study represent peak PRA within 6 months of crossmatch.
For induction immunosuppression, all patients received 30 mg of alemtuzumab as a single intravenous dose given intraoperatively. Patients were premedicated with acetaminophen, diphenhydramine, and 250 mg of intravenous methyl-prednisolone. Maintenance immunosuppression consisted of tacrolimus (Prograf; Astellas, Deerfield, IL) dosed twice daily to achieve a 12-hr trough level of 4 to 7 ng/mL and mycophenolate mofetil (Cellcept; Roche, Nutley, NJ) 500 mg administered orally twice daily. After discharge, CBC, platelet count, electrolytes, creatinine, BUN, and glucose were monitored at least twice weekly for 3 months, weekly once for 3 months, and then twice monthly for the first year. All episodes of increased plasma creatinine level, not clearly attributable to dehydration, or a high tacrolimus level were investigated with a kidney biopsy and read by a single renal pathologist and scored according to Banff criteria (10).
Identification of HLA Antibody
For the virtual crossmatch protocol to be effective, all patients who have HLA antibodies must be identified and the specificities determined. Figure 1(A) outlines the process of HLA screening and identification followed in the HLA laboratory. All patients are screened initially for the presence or absence of HLA class I or II antibody or both using FlowPRA screen beads that are coated with purified HLA class I or II antigens. If the patients screen positive for HLA class I or II or both, Luminex SA beads are then performed to determine the specificity of antibody. This assay detects antibodies to HLA A, B, Cw, DR, DQ, and DP antigens at the allele level. Antigens are assigned as unacceptable if they have a mean fluorescent intensity (MFI) of 2000 or more (shown in the box, Fig. 1B). All antigens with an MFI of 1000 to 2000 will not be assigned as unacceptable but will be marked as “watch” (shown in the oval, Fig. 1B). We have frequently observed that patients who have antibodies in this range will have weakly positive crossmatches; however, this strategy allows discretion for transplantation of the sensitized patient.
Transplantation Rates of Sensitized Recipients
To determine if initiation of a virtual crossmatch protocol had an effect on the rates of kidney transplantation for sensitized recipients, we compared our transplant rates according to PRA with SRTR data from 2006 (Fig. 2). SRTR data for 2006 show that 76% of recipients had PRA 0% to 20%, whereas 14% had PRA 21% to 79% and 5.5% had PRA 80% or higher (4.5% of patients had unknown PRA). In 20 months, since inception of our virtual crossmatch protocol on October 2006, 122 deceased donor kidney transplants were carried out at our center, of which 93 patients (76%) had PRA between 0% and 9%, 15 patients (12%) had PRA between 10% and 79%, and 15 patients (12%) had a PRA of 80% or higher; a significantly higher percent of sensitized patients compared with SRTR data (P=0.0004). Although SRTR data does not report rates of transplantation for patients with PRA of 90% or higher, it is important to note that 9 of 122 (7.4%) of our patients transplanted had PRA between 91% and 100%. To determine the impact of virtual crossmatch on our own center’s rate of transplanting sensitized patients, we reviewed our data from the 12 month period before initiation of our virtual crossmatch protocol. During this time frame, 52 deceased donor transplants were performed, of which three patients (5.8%) had a PRA more than 80% and one patient (1.9%) had a PRA more than 90%. These data are not statistically different from our rates after initiation of virtual crossmatch because of low sample size but are consistent with SRTR data and further suggest that initiation of the virtual crossmatch protocol enhances access of sensitized patients to transplantation.
Outcome of Transplants in Sensitized Recipients
While we were able to achieve a significantly higher rate of transplants into sensitized patients compared with the national average, it was important to analyze our patient outcomes in this group (Table 1). With a median follow-up time of 329 days (range, 30–650 days) 2 of 15 patients (13%) have experienced acute rejection and none of the patients have lost their allografts. Patient 5 had a PRA of 95% and initially did well with a 1-month serum creatinine of 0.9. The patient was noted to have a significant increase in creatinine level during a routine laboratory follow-up and was found to have acute cellular rejection, Banff grade 2A on day 227 posttransplant. The patient was treated with antithymocyte globulin and on most recent follow-up at 18 months posttransplant has a creatinine level of 1.6. Patient 7 had an early antibody-mediated rejection diagnosed on postoperative day 10. Although, her testing 17 months before transplant did not demonstrate a donor-specific antibody, at the time of her rejection episode she had developed a class I donor-specific antibody with a titer of 1:64 ratio. The patient was treated with plasmapheresis and intravenous immunoglobulin, and at most recent follow-up 17 months posttransplant had a creatinine level of 0.9.
Predictive Value of Virtual Crossmatch
To determine the predictive value of our virtual crossmatch protocol, we reviewed the results of all final T- and B-flow cytometry crossmatches carried out during the study period, regardless of whether the kidneys were transplanted (Fig. 3). Of the 136 crossmatches carried out on patients with PRA 0% to 10%, only two patients had positive crossmatches (predictive value 99%) compared with 96% predictive value for patients with PRA 21% to 79% and 89% predictive value for patients with PRA 80% to 100%. Of the two patients with PRA 80% to 100% who had positive final crossmatches, one patient had a recent sensitizing event 3 to 4 weeks before the final crossmatch and as a result developed a donor-specific antibody. The other patient had a weakly positive T-cell crossmatch (median channel shift 70 compared with the cut-off of 57 channels for our center) which may not be a contraindication for transplantation at some other centers.
Sensitized patients are significantly less likely to receive kidneys compared with nonsensitized patients and now make up a significant portion of the national kidney transplant waiting list (11). Thus, efforts to increase kidney allocation to this subset of sensitized patients have been of great interest to the transplant community. One strategy to improve access to sensitized patients was recently described by researchers at Emory University who demonstrated that incorporation of SA bead technology to identify unacceptable antigens into their allocation process resulted in significantly more transplants allocated to sensitized patients (4). Although the Emory group demonstrated excellent long-term survival results in their cohort of sensitized recipients who received allografts, they did not report acute rejection rates. Additionally, they did not show data regarding the predictive value of a virtual crossmatch in their cohort of sensitized patients. Our results are consistent with and extend the Emory group’s observations. In the current study, we show that since inception of the virtual crossmatch protocol, we transplanted significantly more sensitized patients compared with the national average as reported by SRTR. Importantly, we also demonstrate that the acute rejection rate of sensitized patients who received transplants was 13%, with only 1 of 15 patients experiencing an early antibody-mediated rejection that was easily reversed with plasmapheresis and intravenous immunoglobulin. We also demonstrate that the final flow cytometry crossmatch is negative 89% of the time for sensitized patients with PRA more than 80% when it is predicted to be negative by the virtual crossmatch protocol.
A potential limitation of using SA bead technology as a part of the kidney allocation process for sensitized patients is that the detection of donor-specific antibody may not correlate with a positive-flow cytometry crossmatch (12, 13). Thus, sensitized patients could unintentionally be disadvantaged because they would be excluded from kidney offers due to entry of unacceptable antigens that may not be clinically significant. Our approach to minimize the potential impact of this concern is to assign antibodies that are only weakly positive by Luminex as watch and not enter them as unacceptable in the United Network for Organ Sharing (UNOS) database. Follow-up flow crossmatches performed prospectively or retrospectively routinely seem as weakly positive in cases when donors have expressed these watch antigens. Indeed, we believe that if our predictive value of the virtual crossmatch protocol were 100% this may indicate that we are too stringent in our assigning unacceptable antigens. When we do obtain a positive flow crossmatch in the setting of a negatively predicted crossmatch, it allows for individual assessment of the level of risk associated with transplanting patients based on their immunologic profile and sensitization history.
Although we perform comprehensive testing to determine HLA antibody specificity, current UNOS policy does not mandate testing for all identifiable HLA loci. This lack of information could lead to false-negative virtual crossmatches. For example, donors are not routinely typed for DP locus, because this typing currently requires molecular techniques. Interestingly, of our patients with any HLA class II antibody, 33% had detectable levels of DP antibody (MFI>1200) by Luminex assay. Furthermore, a small cohort of patients presented with only DP antibody in the absence of other class II HLA antibody. Initial data indicate that DP antibodies can cause positive B-cell crossmatches; however, further study would be necessary to determine clinical outcomes of such positive crossmatches. Nevertheless, these data suggest that HLA typing of DP should be strongly considered for all organs that are to be shared or shipped for sensitized patients.
Additionally, UNOS does not currently allow entry of allele-specific unacceptable antigens. We believe that this shortcoming could also unnecessarily result in false-negative virtual crossmatch results. For example, a sensitized recipient may be “self” HLA-B*4402 and contain antibody to HLA-B*4403. Currently, the allele-specific B*4403 unacceptable antigen could not be entered into the UNOS computer because it would be recognized as a self HLA-B*44 antigen. This limitation of the current UNOS computer algorithim could thus result in a positive flow crossmatch against a seemingly “zero-mismatch” kidney offer. Of course, it would be necessary to HLA type all donors at the high resolution level to be able to determine allelic level unacceptable antigens.
A limitation of this study is that we have not analyzed the predictive value of a positive virtual crossmatch. Thus, it is possible that in some circumstances even identification of strongly positive donor-specific antibody by Luminex, which we assign as unacceptable, may not be clinically significant and may not result in a positive flow crossmatch (14, 15). These data will be important to explore in future studies. Nevertheless, our data convincingly show that improved antibody identification platforms have increased our level of confidence in the crossmatching process and significantly increases access to transplantation for sensitized patients.
Before initiation of the virtual crossmatch protocol, our standard approach was to perform preliminary screening with anti-human globulin cytotoxicity (AHG-CDC) assay for all patients with PRA levels more than 20%, for every blood group compatible deceased local donor. Historically, we performed approximately 231 preliminary screens per month, at a cost of $40 per AHG-CDC crossmatch. The total cost of performing the preliminary screens for the year before initiation of the virtual crossmatch protocol was $113,000 so the transplant center has recognized a substantial cost savings. Because the cost of antibody screening and identification of antibody specificity was routinely being performed as a part of each patient’s immunologic evaluation before initiation of the virtual crossmatch protocol, these costs have not changed. Another important benefit of the virtual crossmatch protocol is the significant time savings that have been realized. The preliminary screening process would typically add an additional 3 hr to the initial donor histocompatibility testing and thus to the donor allocation process. Despite the preliminary screening process, it was interesting that initiation of the virtual crossmatch protocol resulted in a higher frequency of transplanting sensitized patients. These data suggest that at least some sensitized patients may have been excluded from transplantation by false-positive CDC crossmatches, perhaps due to IgM antibody. In addition, it is also likely that concerns regarding time constraints with the allocation process may have excluded some sensitized patients from final crossmatch consideration. Our virtual crossmatch protocol has now been streamlined such that once the donor HLA typing has been entered into the UNOS computer, kidney allocation is based solely on the patients who appear on the generated match run, thus maximizing efficiency of the process. In conclusion, we have adopted a cost-effective protocol using solid-phase antibody identification techniques to enhance access to kidney transplantation for sensitized patients. If such an approach were applied nationally, we believe that the disparity in allocation for this disadvantaged group would be minimized. We believe strongly that unacceptable antigens should not be assigned to weakly positive antibodies, detected by Luminex and that the predictive value of the virtual crossmatch should be continuously monitored by each transplant center as a “quality-control” index. Additional studies will be necessary to confirm our observations and to more thoroughly explore the potential for assigning clinically insignificant unacceptable antigens with this protocol.
1. Claas FH, van Rood JJ. Transplantation in hyperimmunized patients. Adv Nephrol Necker Hosp
1989; 18: 317.
2. Claas FH, van Rood JJ. The hyperimmunized patient: From sensitization toward transplantation. Transpl Int
1988; 1: 53.
3. Doxiadis II, De Meester J, Smits JM, et al. The impact of special programs for kidney
transplantation of highly sensitized patients in Eurotransplant. Clin Transpl
4. Bray RA, Nolen JD, Larsen C, et al. Transplanting the highly sensitized patient: The Emory algorithm. Am J Transplant
2006; 6: 2307.
5. Bielmann D, Honger G, Lutz D, et al. Pretransplant risk assessment in renal allograft recipients using virtual crossmatching. Am J Transplant
2007; 7: 626.
6. Zangwill S, Ellis T, Stendahl G, et al. Practical application of the virtual crossmatch
. Pediatr Transplant
2007; 11: 650.
7. Zachary AA, Braun WE. Calculation of a predictive value for transplantation. Transplantation
1985; 39: 316.
8. Pei R, Lee JH, Shih NJ, et al. Single human leukocyte antigen flow cytometry beads for accurate identification of human leukocyte antigen antibody specificities. Transplantation
2003; 75: 43.
9. Pei R, Wang G, Tarsitani C, et al. Simultaneous HLA Class I and Class II antibodies screening with flow cytometry. Hum Immunol
1998; 59: 313.
10. Racusen LC, Solez K, Colvin RB, et al. The Banff 97 working classification of renal allograft pathology. Kidney Int
1999; 55: 713.
11. Cecka JM. The OPTN/UNOS Renal Transplant Registry. Clin Transpl
12. Kerman R, Lappin J, Kahan B, et al. The crossmatch
may still be the most clinically relevant histocompatibility test performed. Clin Transpl
13. Kerman RH. Understanding the sensitized patient. Heart Fail Clin
2007; 3: 1.
14. van den Berg-Loonen EM, Billen EV, Voorter CE, et al. Clinical relevance of pretransplant donor-directed antibodies detected by single antigen beads in highly sensitized renal transplant patients. Transplantation
2008; 85: 1086.
15. Gupta A, Iveson V, Varagunam M, et al. Pretransplant donor-specific antibodies in cytotoxic negative crossmatch kidney
transplants: Are they relevant? Transplantation
2008; 85: 1200.