Lan T. Vu, Lee Ann Baxter-Lowe, John Garcia, Marilyn McEnhill, Philip Summers, Ryutaro Hirose, Marsha Lee, Peter G. Stock. HLA-DR Matching in Organ AllocationBalance Between Waiting Time and Rejection in Pediatric Kidney Transplantation. Arch Surg. 2011;146(7):824–829. doi:10.1001/archsurg.2011.147
Author Affiliations: Departments of Surgery (Drs Vu, Baxter-Lowe, Hirose, and Stock; Messrs Garcia and Summers; and Ms McEnhill) and Pediatric Nephrology (Dr Lee), University of California at San Francisco, San Francisco.
Objectives To determine the impact of HLA-DR mismatching on rejection, graft survival, and sensitization in a local allocation system that emphasizes donor quality rather than HLA antigen matching for pediatric patients and to determine the likelihood of finding an appropriate donor based on HLA-DR mismatch.
Design Retrospective cohort study.
Setting A single institution.
Patients A total of 178 patients younger than 21 years who underwent kidney transplantation with daclizumab induction between 1997 and 2006.
Main Outcome Measures The association between HLA-DR mismatching and rejection or graft survival was determined using survival analysis. Sensitization was defined as a posttransplantation panel reactive antibody level greater than 0% in patients with a pretransplantation level of 0%.
Results Median follow-up was 4.1 years (interquartile range, 2.1-6.1 years). One- and 5-year graft survival rates were 97% and 82%, respectively. HLA-DRB1 mismatches were a significant risk factor for rejection; patients with 1- or 2-HLA-DRB1 mismatches had 1.7 times greater odds of rejection than those with 0-HLA-DR mismatches (P = .006). HLA-DRB1 mismatching was not a significant risk factor for either graft failure or sensitization, but history of rejection was an independent predictor of graft failure (hazard ratio, 7.7; P = .01) and sensitization (odds ratio, 9.7; P = .001). Although avoiding HLA-DRB1 mismatching reduces rejection, the probability of finding ABO-matched local donors younger than 35 years without DR mismatches was extremely low.
Conclusion Although avoiding HLA-DRB1 mismatching is beneficial, the likelihood of finding an HLA-DRB1–matched donor should also be considered in donor selection.
Renal allograft failure in children has been associated with several factors, including age, race, donor source, cold ischemia time, primary renal disease, HLA antigen mismatch, and transplantation year.1- 3 Graft survival has improved substantially over the years owing to changes in the immunosuppression regimen, particularly after the addition of cyclosporine and tacrolimus.4 Although avoiding HLA antigen mismatching has been shown to benefit long-term graft survival, it has raised concerns about disadvantaging minority groups, particularly black patients,5,6 and pediatric patients, who have severe growth retardation and other problems when dialysis is prolonged before transplantation.7,8 Currently, only HLA-DR matching is considered in the United Network for Organ Sharing (UNOS) organ allocation system.
In addition, the current United Network for Organ Sharing allocation system assigns preference to children such that avoiding HLA-DR mismatch is no longer a priority. In an effort to shorten deceased donor waiting time, there may be a compensatory increase in the risk of sensitization in children, especially because they are the population that will most likely need at least 1 retransplantation in their lifetimes. The objectives of this study were (1) to determine the impact of HLA-DR mismatching on rejection, graft survival, and sensitization in a local allocation system that emphasizes donor quality rather than HLA antigen matching for pediatric patients and (2) to determine the likelihood of finding an appropriate donor based on HLA-DR mismatch.
This is a retrospective cohort study of pediatric patients (younger than 21 years) who underwent primary kidney transplantation and daclizumab induction therapy between January 1, 1997, and December 31, 2006, at the University of California at San Francisco. During this period, induction with a combined therapy of daclizumab, a humanized anti– interleukin 2 receptor monoclonal antibody, and corticosteroids was standard for pediatric transplant recipients. Maintenance therapy included mycophenolate mofetil and tacrolimus (5-7 ng/mL trough levels after 3 months). This standardized regimen was applied to all primary transplantations and eliminates the confounding factor of different immunosuppression therapies on graft survival. Medical records were reviewed to also ascertain the following patient information: sex, race/ethnicity (white, African American, Hispanic, or other), degree of HLA-DRB1 mismatch, age at transplantation (<13, 13-17, or >17 years), cause of end-stage renal disease (focal sclerosing glomerulosclerosis, glomerulonephritis, structural anomalies [obstructive nephropathy, reflux nephropathy, and renal dysplasia], or other or unknown), pretransplantation dialysis (either peritoneal dialysis or hemodialysis), duration of dialysis (0, <12, 12-24, or >24 months), donor source (deceased or living), development of posttransplantation lymphoproliferative disorder, pretransplantation panel reactive antibody (PRA) level, blood type, and donor's age (in years), sex, and race/ethnicity (white, African American, Hispanic, or other).
Outcome measures included graft failure (initiation of dialysis after transplantation), biopsy-proved acute humoral or cellular rejection, and posttransplantation sensitization. Sensitization was defined as detection of any HLA antigen antibodies after transplantation for recipients with PRA levels of 0% before transplantation. This conservative definition was selected because there is high likelihood that HLA antigen antibodies will continue to increase as patients progress to graft failure, and this higher level of sensitization is the level that is relevant to the likelihood of finding a compatible donor for subsequent transplants. Detection of posttransplantation anti–HLA class I and II antigen antibodies was performed using LABScreen assays per the manufacturer's instructions (One Lambda Inc, Canoga Park, California). Blood samples were collected to test for each patient's current level of the HLA antigen antibodies during the recruitment period. In addition, we reviewed the medical records to obtain information about past PRA levels, which were determined when clinically indicated by the patient's transplant nephrologist. The study was approved by the University of California at San Francisco institutional review board.
Statistical analysis was performed as follows. Graft failure and rejection were treated as time-dependent events, and the Cox proportional hazards regression model was used to determine the association between HLA-DR mismatch and each of those outcomes. Each patient potentially had more than 1 episode of rejection, and, thus, the Prentice, Williams, Petersen (conditional) model was used to analyze the outcome of rejection under the assumption that each patient was not at risk for the second episode of rejection until the first episode had occurred (previous episode of rejection as a risk factor for subsequent episodes). Sensitization, on the other hand, was treated as a time-independent event, and standard covariate adjustment with a logistic regression model was used to determine the association between HLA-DR mismatch and sensitization. The analysis for sensitization included only patients with a PRA level of 0% before transplantation. Other patient- and donor-related variables were included in each of the models if the associated P < .20 in the forward stepwise selection process.
All the analyses were performed using a commercially available software program (STATA, version 9.2; StataCorp LP, College Station, Texas). Two-sided P < .05 was considered statistically significant.
The frequency of 0-HLA-DR mismatch in a local donor pool was determined for each of the 178 patients in this study as a surrogate marker for waiting times. The frequency relates to the estimated probability of finding a 0-HLA-DR–mismatched donor. The frequency of 0-HLA-DR–mismatched donors was determined for 2447 consecutive local donors (California Transplant Donor Network, Oakland) between December 1, 1999, and December 31, 2009. Donors were excluded from the analysis if HLA antigen types were unavailable (n = 53). The frequency of 0-HLA-DR–matched donors was determined for 3 groups: (1) all donors, (2) ABO-matched donors, and (3) ABO-matched donors who were younger than 35 years. A 0-HLA-DR mismatch was defined by the absence of non-self HLA-DR antigens in the donor's molecular or serologic typing. HLA antigen subtypes were considered to be HLA antigen matched with the broad specificity. For example, HLA-DR15 was considered matched with HLA-DR2. HLA-DR103 (n = 15 donors) was not considered a subtype of HLA-DR1. The recipients were stratified by race/ethnicity into African American, Asian, Hispanic, white, and other to illustrate the impact of race/ethnicity on the minimum, median, maximum, and mean frequencies of 0 HLA-DR–mismatched donors.
One hundred seventy-eight patients were identified; 4 patients who required transplant nephrectomy for arterial or venous thrombosis within 24 hours of transplantation were excluded from the analysis. Median follow-up was 49.1 months (interquartile range, 25.7-72.7 months), and median age at time of transplantation was 13.8 years (interquartile range, 10.0-16.6 years). Most recipients were white or Hispanic; only 9.6% (17 of 177) were African American. Structural anomalies were the most common cause of end-stage renal disease; these included obstructive or reflux nephropathy, renal dysplasia, and cortical necrosis. Twenty-two percent of the patients (38 of 177) received preemptive transplantation (before the initiation of dialysis); among patients who required either peritoneal dialysis or hemodialysis before transplantation, the median waiting time for a transplant was 11.8 months (interquartile range, 6.0-21.2 months) (Table 1).
One year after transplantation, rejection occurred in 35% of the patients; at 5 years, the frequency of rejection was 55%. Patients with 1- or 2-HLA-DRB1 mismatches had 1.7 times greater odds of rejection than those with 0-HLA-DRB1 mismatches (95% confidence interval [CI], 1.1-2.7; P = .006). Other risk factors associated with rejection included an increasing number of total HLA antigen mismatches, older age, development of posttransplantation lymphoproliferative disorder, pretransplantation dialysis for longer than 24 months, recurrence of primary renal disease, and recipient's and donor's race/ethnicity (Table 2).
The 1- and 5-year graft survival rates for this cohort were 97% and 82%, respectively. In the multivariate Cox proportional hazards regression model, the degree of HLA-DRB1 mismatches did not significantly affect graft failure. However, patients with a history of rejection had 7.7 times greater odds of graft failure than those who never had an episode of rejection (95% CI, 1.6-37.7; P = .01). Other risk factors for graft failure were older age and the development of posttransplantation lymphoproliferative disorder (Table 3).
Posttransplantation serum samples were obtained from 85 of the 124 patients (68.5%) who had a pretransplantation PRA of 0%. Of the patients who had posttransplantation serum samples, 24.7% (21 of 85) had graft failure, and all the patients who were lost to follow-up still had functional grafts at their last clinic visit (Table 4). The frequency of sensitization in this cohort was 57.6% (49 of 85); 59.2% of the sensitized patients (29 of 49) had functional grafts. Again, degree of HLA-DRB1 mismatch was not a statistically significant (P < .05) risk factor for sensitization in this population. However, patients who had at least 1 episode of rejection had 9.7 times greater odds of becoming sensitized than those who had never rejected their graft (95% CI, 2.5-37.6; P = .001) (Table 5).
Because recipients with 1- or 2-HLA-DRB1 mismatches had 1.7 times greater odds of rejection than those with 0-HLA-DR mismatches, the frequency of finding a 0-HLA-DR–mismatched local donor was determined. The frequency of 0-HLA-DR–mismatched local donors who were younger than 35 years and ABO identical was very low for the study participants, regardless of race/ethnicity (Table 6).
This single-center study demonstrated that HLA-DRB1 mismatching increased the risk of allograft rejection by approximately 70% in children. In addition, rejection was one of the most important predictors of graft failure and sensitization. Sensitization is a particular concern for pediatric transplant recipients because most patients will eventually need a subsequent allograft(s), and sensitization will affect the frequency of compatible donors and outcomes of subsequent transplantations. Avoiding HLA-DRB1 mismatching may decrease the incidence of rejection, but it would increase waiting time, which is also deleterious. The estimated frequency of 0-HLA-DRB1–mismatched, ABO-compatible donors was less than 2 per 100 local donors for most of the patients.
The strengths of this study include the quality of follow-up data, the ability to obtain blood samples to detect PRA in patients who have functional grafts, and elimination of the confounding effect of different immunosuppression regimens. A recent registry study by Gritsch et al9 demonstrated that 0-HLA-DR–mismatched kidneys had statistically significant comparable 5-year graft survival as 1- and 2-HLA-DR–mismatched kidneys in primary pediatric kidney transplant recipients in the United States. However, in registry studies, the impact of HLA antigen mismatching is often not detectable until 10 years after transplantation. Using registry data, Gritsch et al9 reported that HLA antigen disparity did not result in a detectable difference in the odds of developing a PRA level greater than 30% at the time of second wait listing. However, registry data are affected by substantial variation in reporting of PRA levels, and this study could not assess HLA antigen antibodies in recipients who had not yet lost graft function. This center-specific study agreed with the registry study in that graft failure was not statistically significantly different between 0-HLA-DR–mismatched and 1- or 2-HLA-DRB1–mismatched recipients, whereas the rejection rate was statistically significantly greater in those who had 2-HLA-DRB1–mismatched donors. In addition, both studies detected a “dose effect” when more HLA antigens were mismatched between recipient and donor.
Controversy remains regarding the importance of HLA antigen matching in the current era of immunosuppression therapy. Data from the UNOS and the United States Renal Data System from 1994 to 1998 suggested a diminishing significance of HLA antigen matching in kidney transplantation with time.10 Several studies suggest that HLA-DRB1 mismatching is particularly deleterious, perhaps because HLA-DRB1 is a marker of closely linked HLA antigen loci, which were not considered (HLA-DRB3, DRB4, and DRB5 and HLA-DQ). Roberts et al6 concluded that avoiding HLA-DRB1 mismatching alone would increase transplantation in nonwhite individuals by 6.3% while decreasing the numbers in white patients by only 4.0%, with a resultant 2% decrease in graft survival. In addition, a small retrospective review11 of 88 pediatric patients who underwent first cadaveric renal transplantation supported the importance of only HLA-DRB1 mismatching on short- and long-term graft survival.
Delaying transplantation can be deleterious. Increased time on the waiting list for adults with end-stage renal disease has been associated with dialysis-related mortality, such as cardiovascular events. In addition, increased duration of dialysis before transplantation has been associated with poor graft outcomes. Patients who had been undergoing dialysis at least 6 months before transplantation had an increased risk of graft failure compared with those who had preemptive transplantation.12 However, it remains unclear whether preemptive transplantation would benefit pediatric transplantation patients owing to conflicting reports in the literature.1,13- 15 The most consistent finding for children with chronic renal disease has been that long-term dialysis and chronic renal disease are associated with growth failure and short stature. Moderate and severe growth failure was demonstrated to increase the risk of death 2- to 3-fold in pediatric patients with renal failure.7,16 In addition, the present study demonstrated that patients who were undergoing dialysis for longer than 2 years before transplantation had a higher risk of graft rejection.
Based on the results of this relatively large study of pediatric transplant recipients, we recommend changes in the deceased kidney allocation system in the United States to improve graft outcomes and decrease waiting time for children. We recommend that the frequency of 0-HLA-DRB1–mismatched donors be calculated for each patient on the waiting list. For those who may have a higher probability, waiting for a 0-HLA-DRB1 mismatch will benefit these patients in terms of decreased risk of rejection and, thus, sensitization and graft failure. However, those who have an exceedingly low probability, such as ethnic minorities, should undergo transplantation regardless of HLA-DRB1 mismatch. For these patients, it might be beneficial to use approaches to select HLA-DRB1–mismatched donors who have a lower likelihood of mismatching for linked HLA class II antigen loci (HLA-DRB3, DRB4, and DRB5 and HLA-DQ). Ultimately, for each patient, the potential benefit of avoiding HLA-DRB1 mismatching should be balanced by the detrimental effects of increased waiting time.
Although this study is relatively large for pediatric patients from a single institution, it has several limitations. We studied only patients at the University of California at San Francisco to examine the role of HLA antigen mismatching in an allocation system that does not place priority based on HLA antigen mismatch. Therefore, the resultant smaller sample size may have minimized the power to detect potential associations. Other confounding factors, such as medical nonadherence, may have contributed to the outcomes of rejection, graft failure, and sensitization and were not captured in this study. Age was a stronger predictor of all 3 outcomes and may have served as a surrogate marker for nonadherence. When we reviewed the study group, we found nonadherence of 27% based on subjective observation in clinic notes. Nonadherence has been well documented as a risk factor for poor outcomes in children.17- 19 However, we chose to exclude this variable in the analysis because medical nonadherence was difficult to verify objectively and retrospectively.
The diagnosis of rejection and sensitization were based on retrospective medical record review and available blood samples. The frequency of rejection may have been underestimated. Humoral or antibody-mediated rejection has been described only recently,20 and, therefore, patients in the early study period who had acute tubular necrosis on biopsy may have been misdiagnosed. In addition, we obtained posttransplantation blood samples only from patients who were currently seen at the University of California at San Francisco Pediatric Renal Clinic or those who had graft failure. Among patients who had a pretransplantation PRA level of 0%, the only difference between those who were lost to follow-up and those who had blood samples for posttransplantation PRA testing was that all the patients who were lost to follow-up still have functional grafts and, therefore, are less likely to be sensitized. Thus, the frequency of sensitization may have been overestimated, but the point estimates for association between different variables and sensitization should have been unaffected.
Sensitization is a particular concern for pediatric patients, who are likely to require multiple grafts during their lifetime. Leffell et al5 reported that the likelihood and extent of sensitization are increased as the number of HLA antigen mismatches in the donor increases. In this study, 78% of the recipients received grafts from donors who had 3 or more HLA antigen disparities, consistent with detection of HLA antigen antibodies in a large percentage of the cohort. Many investigations1,2,4,6,10,11,13 have shown a strong relationship between HLA antigen antibodies and graft loss. Furthermore, after the graft has been lost, the presence of HLA antigen antibodies makes it more difficult to find a compatible donor for retransplantation, and subsequent grafts also have an increased likelihood of failure. Perhaps some of these problems will be mitigated by new approaches to reduce HLA antigen antibodies before the graft is lost and by desensitization protocols to reduce antibody levels before subsequent transplantations.
In conclusion, this study showed that HLA-DRB1 mismatch was a risk factor for rejection and that rejection was a strong predictor of graft failure and sensitization in children. However, the probability of finding a 0-HLA-DRB1–mismatched donor for this cohort was low for all the patients, with the chance of getting a local match within a year extremely unlikely. For Asian patients, the likelihood was negligible. Based on the current allocation system, most pediatric recipients would benefit more by minimizing dialysis time and accepting an HLA-DR–mismatched kidney. Although the current allocation algorithm is principally restricted to local donors, expansion to a system of regional allocation would provide the opportunity for better matching for these young recipients to maximize outcomes and prevent sensitization. Based on the high incidence of rejection and sensitization seen with HLA-DR–mismatched kidneys observed in this series, we advocate expansion of the current artificial boundaries of donor pools to facilitate better matching and outcomes for young recipients. In the absence of such a policy change, it would be unrealistic or inappropriate to recommend HLA-DR matching, and transplantation centers should proceed with the transplantation of HLA-DR–mismatched kidneys to minimize the morbidities associated with dialysis in the pediatric population.
Correspondence: Peter G. Stock, MD, PhD, Department of Surgery, University of California at San Francisco, 505 Parnassus Ave, Moffitt Bldg M884, Box 0780, San Francisco, CA 94143-0780 (firstname.lastname@example.org).
Accepted for Publication: May 29, 2010.
Author Contributions:Study concept and design: Vu, Baxter-Lowe, McEnhill, and Stock. Acquisition of data: Vu, Baxter-Lowe, Garcia, and McEnhill. Analysis and interpretation of data: Vu, Baxter-Lowe, Garcia, Summers, Hirose, Lee, and Stock. Drafting of the manuscript: Vu, Summers, and Stock. Critical revision of the manuscript for important intellectual content: Vu, Baxter-Lowe, Garcia, McEnhill, Hirose, Lee, and Stock. Statistical analysis: Vu, Garcia, Summers, Hirose, and Lee. Obtained funding: Baxter-Lowe. Administrative, technical, and material support: Baxter-Lowe, Garcia, McEnhill, and Stock. Study supervision: Stock.
Financial Disclosure: None reported.
Previous Presentation: This study was presented at the 81st Annual Meeting of the Pacific Coast Surgical Association; February 15, 2010; Kapalua, Hawaii.