Influence of HLA Matching on Thoracic Transplant Outcomes
An Analysis From the UNOS/ISHLT* Thoracic Registry
Background The benefit of matching donor organs and recipients for HLAs has been well established in renal transplantation and has been suggested for thoracic organ transplantation. To determine the effect of HLA matching in cardiac and single-lung transplantation, the following study was performed.
Methods and Results Using the joint Thoracic Transplant Registry of the United Network for Organ Sharing and the International Society for Heart and Lung Transplantation, all adult primary heart and single-lung transplant procedures performed in the United States from October 1987 through December 1993 were analyzed to determine the effects of HLA matching on transplant mortality. Both total HLA matches and matches at individual HLA loci were considered. Including HLA matching, 16 potential risk factors for heart transplant outcome and 16 potential risk factors for lung transplant outcome were subjected to multivariate analysis. A total of 10 752 heart transplants and 1239 lung transplants were included in the independent analyses for each organ. For heart transplantation, there was a progressive reduction in risk for greater matching (1 or 2 matches: risk ratio, 0.83; 3 matches: risk ratio, 0.67; 4 to 6 matches: risk ratio, 0.59; all P≤.01). The primary benefit of matching appeared to be at the A and DR loci (risk ratios, 0.87 and 0.79, respectively; P<.001). For lung transplantation, any matching had an independent positive effect on outcome; however, the relationship between numbers of HLA matches and relative risk was not present and ranged from risk ratios of 0.71 to 0.87 (P=.01 and P=.47, respectively). In this analysis, only matching at the A locus appeared to statistically influence outcome (risk ratio, 0.76; P=.01).
Conclusions These data demonstrate that HLA matching independently impacts survival in both heart and single-lung transplantation.
The benefit in survival afforded by HLA matching has been well established in kidney transplantation,1 2 and sharing of kidneys based on high degrees of histocompatibility matching is now established national policy.3 Because of poor preservation of thoracic organs and the very short (4 to 8 hours) allowable cold ischemic times, prospective HLA matching is not currently possible. The effect of HLA matching in thoracic organs has been difficult to ascertain, given the very few recipients overall who fortuitously receive histocompatible organs. Furthermore, the total number of thoracic organs transplanted at a single center is small, making single-center analysis of HLA matching impossible. Accordingly, the Combined Thoracic Registry of the United Network for Organ Sharing and the International Society for Heart and Lung Transplantation (UNOS/ISHLT) was analyzed for the effects of HLA matching on outcome after heart and single-lung transplantation.
All adults >18 years old undergoing either heart or single-lung first-time transplants from October 1987 through December 1993 from all centers in the United States were included in the analysis. These data were reported to the Registry, individual data entry forms were then returned to the individual transplant centers for verification, and data on transplants performed through December 1991 were reverified by the individual centers as part of the UNOS center-specific survival project.4 Outcomes (alive or dead) and date of follow-up were based solely on receipt of follow-up data verifying the individual patient's status, and patients lost to follow-up (<1%) were censured at the time of most recent data submission.
Because of the small numbers of organ recipients who had high HLA concordance with their donors, patients with the highest matches (four to six matches in heart and three to six matches in lungs) were grouped for analysis. Likewise, when the effect of a specific HLA locus was analyzed, the group with one or two matches at that locus was compared with the group with no matches at that locus. Survival curves were generated by the Kaplan-Meier method, and differences in survival curves were analyzed by the log-rank test.
Multivariate analysis to determine independence of HLA matching as a risk factor was performed by proportional-hazards regression. A total of 16 potential risk factors for heart transplantation and 16 potential risk factors for lung transplantation were included in the model and are shown in Table 1⇓. Complete HLA typing was not available for every donor/recipient pair in the data (≈24% of heart transplants and 21% of lung transplants). To ensure against a bias based on missing HLA data, covariates were included for missing HLA match in each analysis. To control for a possible center-level reporting bias, the data were reanalyzed with a proportional-hazards regression stratified by transplant center. Results similar to the unstratified analysis were obtained. Significance was considered present at a value of P<.05.
The demographics for the 10 752 patients analyzed are presented in Table 2⇓ and Fig 1⇓. The majority were white males in the fifth and sixth decades of life, and the two most common indications for heart transplantation were ischemic heart disease and cardiomyopathy. Table 3⇓ shows the number of patients matched at each level of the six total possible loci as well as the numbers of patients matched for HLA-A, HLA-B, and HLA-DR.
Fig 2⇓ demonstrates the Kaplan-Meier survival curves for 3-year survival for each level of HLA matching. There was a statistically significant reduction in survival (P<.001) for those with the poorest match in this univariate analysis. Table 4⇓ demonstrates the multivariate proportional-hazards regression analysis for those risk factors that were not related to histocompatibility but were statistically significant as predictors for 1-year survival. There were no differences in this analysis when a proportional-hazards analysis stratified by transplant center was performed, negating a center-specific effect not accounted for by the included covariates. Table 5⇓ demonstrates the impact of HLA matching in this analysis. The adjusted risk of mortality in the missing HLA group was marginally better than that for the reference HLA category (0 matches) with a risk ratio of 0.90, P=.046. This reduction in risk was not even as great as that seen for patients and donors with only one match, indicating that the patients with missing data were most similar to the nonmatched reference group. There is a progressive and independent reduction in mortality with increasing levels of HLA matching in this population. The impact of HLA matching after heart transplantation appears to lie largely in the HLA-A and DR loci, with HLA-B matching not having an independently significant effect on outcome.
The demographics for the 1239 patients analyzed are presented in Table 6⇓ and Fig 3⇓. There was a more equal split between men and women in this cohort. As with heart transplant recipients, the majority of patients were in the fifth and sixth decades of life, and the common indications for lung transplantation were emphysema, interstitial pulmonary fibrosis, α1-antitrypsin deficiency, and primary pulmonary hypertension. Table 7⇓ shows the number of patients matched at each level of the six total possible loci as well as the numbers of patients matched for HLA-A, HLA-B, and DR. There were very few patients with high degrees of matching (no patients with five or six HLA matches), which probably influenced the results.
Fig 4⇓ demonstrates the Kaplan-Meier survival curves for 3-year survival for each level of HLA matching. Although there was a marginally statistically significant reduction in survival for patients with no matches (P=.07), there was no trend with more than 0 matches for improved survival. This is further confirmed by the multivariate analysis (Table 8⇓), which shows a statistically significant reduction in risk with any degree of HLA matching but no progressive improvement with increasing levels of matching. The group of patients with missing HLA data was not different from the reference group (0 matches). Interestingly, as in heart transplantation, matching at the A locus was again a statistically significant positive factor. In this case, DR matching showed a reduction in risk but was not statistically significant, probably because of the smaller numbers of patients in each group. Table 9⇓ demonstrates other non–histocompatibility-related risk factors that were statistically significant predictors of 1-year outcome after lung transplantation. There did not appear to be an independent center effect.
The analyses presented here demonstrate that HLA matching clearly has an independent impact on adult, first-time heart transplant survival even when the majority of important risk factors for early to intermediate outcome are accounted for. HLA matching for adult, first-time lung transplantation likewise appears to affect outcome; however, these latter analyses are hampered by relatively low numbers of recipients, especially in the highly matched groups. Unfortunately, given that these data are generated from the largest, most complete and verified transplantation database in existence, other sources of accurate lung transplant data with larger numbers are not available. It is interesting that in both thoracic organs, matching at the HLA-A and DR loci appears to be responsible for the majority of the effects seen in total matching.
The first study investigating HLA matching in heart transplantation in a large number of patients was reported by Opelz and Wujciak.5 In this voluntary international registry containing approximately 8300 patients from 24 countries, the authors demonstrated an independent effect of HLA matching on survival. Only a small number of non-HLA variables were included in this analysis, however. In addition to the larger study, a number of smaller, either single-center or multicenter analyses of the effects of HLA matching on rejection and/or survival have been performed, most using univariate analyses and containing small numbers of patients.6 7 8 9 10 11 12 A more recent study by De Mattos and colleagues,13 using high-confidence-level serological DR typing, demonstrated a reduction in acute rejection and an increase in graft survival in patients with 0 or 1 DR mismatch compared with those with 2 mismatches. Again, however, analysis was multiple comparison and univariate.
We are aware of only one study investigating the role of HLA matching in lung transplantation. Iwaki and colleagues,14 using data from Pittsburgh on 74 either single- or double-lung allograft recipients, demonstrated that DR matching appeared to predict outcome. Those patients with the poorest outcomes had two mismatches at the HLA-A locus, and there was no consistency to matching at the HLA-B locus. The patients with the greatest number of total HLA mismatches had the poorest outcome at 6 months. As with the smaller heart transplant studies, no other risk factors were taken into account in this study.
The present study clearly has limitations. First, this is a retrospective analysis of a multicenter registry. The primary issues include the following: (1) Not all possible risk factors are available in the registry for analysis as covariates. (2) Although many of the data elements (including all of the non-HLA risk factors) for each patient have been subjected to a repeat verification step, HLA typing on thoracic recipients was taken as originally submitted by the transplant centers. (3) HLA typing on thoracic organ recipients is not compulsory, and hence, some portion of HLA data was missing for between 20% and 30% of the recipients. (4) HLA typing from center to center is variable; hence, the accuracy of the actual typing cannot be confirmed.
The strengths of this study are that the UNOS/ISHLT Registry is the largest thoracic transplant registry in existence. Furthermore, its use is compulsory for all transplant centers performing thoracic organ transplantation in the United States, and it therefore includes all patients from all centers in this country. Data are subjected to verification and auditing procedures, and the error rates, based on the UNOS center-specific survival study,4 are <2% overall. Finally, follow-up in this patient population is virtually complete.
Unequivocal data are now available demonstrating the benefits of HLA matching in heart transplantation. It is anticipated that, with larger numbers of lung transplant procedures performed, similar data will ultimately be available for recipients undergoing this procedure. Unfortunately, at the present time, the transplant community is unable to put this information into clinical practice, because of the relatively short cold ischemic times that are allowable for thoracic organs. With improved organ preservation, the ability to share thoracic organs over greater distances may become a reality, thereby allowing HLA matching. If alternatives to thoracic transplantation, such as mechanical ventricular assist devices for cardiac patients, become readily available, then organ allocation priorities will most likely be focused on the optimal survival of the transplanted organ.
- Received August 7, 1995.
- Revision received December 28, 1995.
- Accepted January 2, 1996.
- Copyright © 1996 by American Heart Association
Takemoto S, Gjertson DW, Terasaki PI. HLA matching: a comparison of conventional and molecular techniques. In: Terisaki PI, Cecka JM, eds. Clinical Transplants 1992. 8th ed. Los Angeles, Calif: UCLA Press; 1993:413-434.
Pirsch JD, D'Alessandro AM, Sollinger HW, Hoffmann RM, Roecker E, Voss BJ, Lorentzen D, Knechtle SJ, Reed A, Kalayoglu M, Belzer FO. The effect of donor age, recipient age and HLA match on immunologic graft survival in cadaver renal transplant recipients. Transplantation. 1992;53:55-59.
McClelland J, Steidler K, Cecka JM, Barba L, Cohen J, Dainko E, Danovitch G, Darras F, Ettenger R, Hardy B, Lieberman E, Mackett T, Martin D, Mendez R, Mendez R, Rosenthal JT, Ruzics E, Tertzakian G, Teichman S, Walstrom E. Kidney allocation under the UNOS point system: an update. In: Terisaki PI, Cecka JM, eds. Clinical Transplants 1992. 8th ed. Los Angeles, Calif: UCLA Press; 1993:405-413.
United Network for Organ Sharing. 1994 Report of Center-Specific Graft and Patient Survival Rates: Executive Summary. Washington, DC: US Dept of Health and Human Services; 1994:1-6.
Khaghani A, Yacoub M, McCloskey D, Awad J, Burden M, Fitzgerald M, Hawes R, Holmes J, Smith J, Banner N. The influence of HLA matching, donor/recipient sex and incidence of acute rejection on survival in cardiac allograft recipients receiving cyclosporine A and azathioprine. Transplant Proc. 1988;21:687-688.