For pediatric patients with end-stage renal disease who require a kidney transplant, selection of an optimal donor is particularly important because these patients are young and hope to live with a functioning transplant well into adulthood, and indeed for the remainder of their lives. Transplants from parental donors have been shown to yield very good success rates and parents are therefore preferred donors for children in need of a kidney transplant. For a variety of reasons, parental donor transplantation is not always possible, and alternative donor sources therefore may have to be sought. Other closely or distantly related family members or altruistic unrelated volunteers are potential donor kidney sources, in addition to organs from deceased donors (DDs). The latter are usually searched for by listing potential recipients on a national transplant waiting list, a procedure that is associated with a waiting period which commonly depends on the availability of a donor with a reasonably good HLA match. Kidney allocation algorithms in many countries have been adjusted to give priority to pediatric recipients over adults, so that the waiting time for children is kept as short as possible. Nevertheless, the availability of a living donor (LD) carries an important advantage because the often-prolonged waiting time for an HLA well-matched DD kidney can be avoided during the critical period of body growth in children. Much shorter ischemic kidney preservation with living as compared with deceased donation is considered another important advantage of transplantation of kidneys from LDs. Relatives, even if blood related, are not necessarily HLA well matched, and with an unrelated LD, the likelihood that a given recipient will be a good HLA match is very small. It is therefore of considerable interest whether the outcome of a pediatric kidney transplant from a LD who is poorly HLA matched can be expected to be as good or better than one from an HLA well matched DD. This important question was addressed in a recent publication based on data from the UK transplant registry. In the present study, we show results that disagree with the conclusion reached by our UK colleagues, and we discuss likely reasons for this discrepancy.
MATERIALS AND METHODS
The analysis was based on the data of the international Collaborative Transplant Study (CTS).1 First pediatric kidney transplants performed from 2000 to 2015 were analyzed. The recipients were aged 0 to 17 years and LDs were 18 years or older at the time of transplantation. All available data on pediatric transplants performed during the same period from DDs with 0 to 1 HLA-A+B+DR mismatches (MMs) were analyzed for comparison with grafts from LDs with 4 to 6 HLA-A+B+DR MMs. HLA typing for the HLA-A, -B, and -DR loci was performed at laboratories of the participating centers and reported to the CTS study center shortly after transplantation when the transplant was registered for the study. For transplant waiting lists and donor kidney allocation, organ allocation agencies use HLA antigen specificities determined by low-resolution HLA typing. Low-resolution HLA specificities were therefore used also for the calculation of HLA MMs in this study. For analysis of HLA MMs on graft survival, the numbers of MMs at the HLA-A, -B, and -DR loci were added. Clinical outcome data were recorded at 3, 6, and 12 months, and annually thereafter. Graft survival rates were computed using the Kaplan-Meier method. Mantel Cox log rank test was used for trend analysis of the impact of HLA MMs on graft survival. Cox multivariable regression analysis was performed to account for the possible influence of the following confounders: geographical origin (continent), year of transplant, recipient and donor age, sex and race, preformed panel reactive antibodies, time on dialysis before transplantation, general evaluation of patient as a candidate for transplantation at time of transplant, original disease leading to end-stage renal failure, donor relationship, and immunosuppressive medication (type of calcineurin inhibitor, type of antimetabolite, antibody induction). The software package IBM SPSS Statistics version 24.0 (SPSS Inc, IBM Corporation, Somers, NY) was used. P values below 0.05 were considered significant.
Data from 3627 LD kidney transplants into pediatric patients performed during the years 2000 to 2015 were analyzed. The geographical distribution of recipients was: 36% Europe, 35% Asia, 12% Latin America, 9% Australia/New Zealand and 8% North America. Of the live kidney donors 74% were parents; 7% unrelated; 7% grandparents, aunts, uncles, or cousins; 6% siblings; and 6% were reported as related but the degree of relationship was not specified. Demographics are detailed in Table S1, SDC, http://links.lww.com/TP/B448.
The influence of HLA MMs on survival of these transplants during 10 years of follow-up is depicted in Figure 1. The correlation of number of MMs with graft survival was statistically significant (log rank with trend, P < 0.001). Three main groups could be discerned with respect to their graft survival rates: the HLA 0 MM group, the HLA 1 to 3 MMs, and the HLA 4 to 6 MM transplants (Figure 1). Using the 0 MM grafts as reference, Cox multivariable regression analysis considering the confounders listed under Materials and Methods showed a 2.20-fold higher likelihood of graft failure for the 1 to 3 HLA MM group: hazard ratio, 2.20 (95% confidence interval, 1.39-3.49; P < 0.001) and a 3.91-fold risk of failure for the 4 to 6 HLA MM group, hazard ratio, 3.91 (95% confidence interval, 2.37-6.45; P < 0.001).
Because these results disagreed with the conclusion reached by Marlais et al,2 we performed a second analysis in which we grouped the CTS transplants into the match levels 1 to 4 per the definition of the UK 2006 National Kidney Allocation Scheme3,4 exactly as done by Marlais et al. The result is illustrated in Figure 2. In this analysis, level 1 is identical with our definition of 0 HLA-A+B+DR MMs. Levels 2 and 3 combined include nearly all 1 to 3 HLA-A+B+DR MMs as defined in our analysis (but 56 of the level 2 and level 3 transplants had 4 HLA-A+B+DR MMs), and level 4 corresponds mainly to the 4 to 6 MM group of our analysis (but 41 of these grafts had only 2-3 HLA-A+B+DR MMs). The correlation of match levels 1 to 4 with graft survival was statistically significant (log rank with trend, P < 0.001) (Figure 2).
The important question whether HLA poorly matched LD transplants do as well as HLA well-matched grafts from DDs was analyzed by comparing the graft survival rates of the LD 4 to 6 HLA-A+B+DR MM transplants with the DD 0 to 1 HLA-A+B+DR MM transplants performed during the same period. The 0 to 1 MM group was chosen based on the results of an analysis of deceased-donor transplants in which 0 to 1 MM grafts were found to have equivalent outcome, which was superior to that of transplants with more MMs (Figure S1, SDC, http://links.lww.com/TP/B448). The resulting 10-year graft survival rates are shown in Figure 3. HLA well-matched grafts (0–1 MMs) from DDs survived at a significantly higher rate than HLA poorly matched (4–6 MMs) LD grafts (log rank P = 0.006). Of the 364 LD transplants with 4 to 6 HLA MMs, 149 were from blood-related donors, whereas 215 were from blood-unrelated donors. The graft survival rates of these 2 groups of transplants were virtually identical (log rank, P = 0.99, not shown). The survival rate of 0 to 1 MM DD grafts was not better than that of 1 to 3 MM LD grafts (P = 0.33, Figure 3). To ascertain that immunological graft loss and not patient death was responsible for the observed effect, rates of death-censored graft survival (an approximation of losses due to immunological causes) and patient survival were computed in addition (Figure S2, SDC, http://links.lww.com/TP/B448).
In pediatric kidney transplantation, it is of importance to know whether HLA matching influences the outcome of grafts from LDs, and whether HLA poorly matched transplants from LDs can be expected to do as well as HLA well-matched DD transplants. Marlais et al2 recently addressed these questions and concluded that HLA matching neither had a significant impact on the survival rate of pediatric living nor on DD transplants, and that all pediatric kidney grafts from LDs, regardless of HLA match, did better than even the best-matched DD transplants. If confirmed, these findings would have important practical implications. If several relatives were available as potential donors, the degree of HLA match with the recipient could be disregarded. Moreover, any relative regardless of HLA match, as well as any unrelated LD, always would be a preferred donor compared with the option of transplantation from a DD. This would provide strong motivation for HLA poorly matched relatives or altruistic unrelated LDs to provide donor kidneys to children with end-stage renal disease. It is important to bear in mind that by chance alone, due to the extensive polymorphism of the HLA system, the likelihood that a kidney donated by an unrelated altruistic donor will be poorly HLA matched is very high.
Analysis of the data contained in the CTS database yielded results that disagreed with the conclusion reached by Marlais et al. We found a strong and statistically significant impact of HLA-A+B+DR MMs on the survival of pediatric transplants from LDs and, furthermore, a significantly better transplant outcome with kidneys from DDs with 0 to 1 HLA MMs as compared to transplants from LDs with 4 to 6 MMs. Our main conclusion, therefore, is that LD transplants with more than 3 HLA-A + B + DR MMs should be performed only under certain circumstances, for example, when the potential recipient possesses such a rare HLA phenotype that the likelihood of finding a well-matched (0-1 HLA MM) DD kidney is so small that the relatively low survival rate of a poor LD match is deemed acceptable. The likelihood of finding a well-matched DD within a given period on the waiting list, also termed “matchability,” can be readily calculated.5
A probable reason for the results of the 2 studies to differ is the availability of a sufficiently large data material to analyze the relatively rare extremes, the perfect, and the very poor HLA matches. There were only 9 very poor level 4 matches and 31 perfect level 1 matches in the analysis of Marlais et al.2 The authors combined the level 1 and level 2 grafts and the level 3 and level 4 grafts to obtain groups large enough for statistical analysis. Per our analysis using the level 1 to 4 gradation, however, combining the small number of level 1 with the much larger number of level 2 grafts concealed the far superior survival of level 1 transplants, and combining the small level 4 group with the much larger level 3 group concealed the far inferior outcome of level 4 grafts (Figure 2). It would have been better to combine the level 2 and level 3 transplants, which show similar survival rates, and compare with level 1 and level 4 grafts, but, as stated, the patient numbers in these latter 2 categories were very small in the analysis of Marlais et al.2
Our present results are supported in that we have previously found a statistically significant effect of HLA matching in an analysis of 9209 pediatric transplants from DDs.6 The cited publication by Su et al7 claiming a diminishing effect of HLA matching with modern immunosuppression has been contradicted by an analysis of the CTS data8 as well as by the recent analysis of Williams et al9 based on the complete United States registry data. Additionally, a recent analysis of United States registry data on HLA matching in adult LD transplantation10 also supports our present findings in pediatric LD transplants.
We conclude that kidney transplantation from LDs is an excellent option for pediatric patients with end-stage renal failure, provided the number of HLA-A+B+DR MMs does not exceed 3. Transplantation of LD kidneys with 4 to 6 HLA-A+B+DR MMs showed inferior graft survival, both compared to LD grafts with fewer than 4 MMs and DD grafts with 0 to 1 MMs. Donation from a parental donor is not affected by this conclusion because, per the law of genetic inheritance, the number of HLA-A+B+DR MMs between child and parent cannot exceed 3. Other potential related donors may have more than 3 MMs, and HLA typing will readily provide the information to assess compatibility. Much larger data sets will be required to study whether a MM at 1 HLA locus (A or B or DR) is more harmful than a MM at another. We were unable to identify such differences among MMs and therefore used the conventional method of simply adding up the numbers of HLA-A, -B, -DR MMs for analysis.
LD transplantation is associated with a very short waiting time, an advantage that is particularly relevant for children during the main period of growth. This advantage must be weighed against the significantly lower success rate of 4 to 6 HLA MM grafts from LDs as compared with that of 0 to 1 HLA MM grafts from DDs. A reasonable donor search strategy for a pediatric transplant candidate would seem to first consider the availability of a preferred LD with fewer than 4 HLA MMs. If no such donor is available, evaluation of the potential recipient's HLA phenotype and the likelihood of obtaining a 0 to 1 MM kidney from a DD within an acceptable period would seem appropriate. If none of these options is available, transplantation of an LD kidney with more than 3 MMs or a DD kidney with more than 1 MM would be acceptable practical options. Generally, because children may need another transplant in their lives when a first transplant fails, HLA mismatching should be kept to a minimum to avoid antibody formation against HLA which might result in many unacceptable HLA MMs for the next transplant.
The authors thank the following 147 transplant centers and staff for their support: Aachen, Adelaide, Amsterdam, Ankara, Antalya, Auckland, Bangkok, Barcelona, Bari, Basel, Beilinson, Belfast, Belo Horizonte, Berlin, Bern, Birmingham, Bochum, Bogota, Bologna, Bonn, Botucatu, Brisbane, Bristol, Brussels, Budapest, Buenos-Aires, Cagliari, Cairo, Cambridge, Cape Town, Caracas, Cardiff, Catania, Christchurch, Cleveland, Cologne, Coventry, Dallas, Dublin, Duesseldorf, Edinburgh, Edmonton, Erlangen-Nuernberg, Erzurum, Essen, Florence, Frankfurt, Freiburg, Genoa, Giessen, Glasgow, Goteborg, Grand Rapids, Halle, Hamilton, Hannover, Heidelberg (AUS), Heidelberg (D), Helsinki, Hong Kong, Innsbruck, Istanbul, Izmir, Jena, Kaiserslautern, Kansas City, Karachi, Korea, Leeds, Leicester, Leuven, Liege, Lille, Lima, Linz, Loma Linda, London, Luebeck, Lyon, Maastricht, Madrid, Mainz, Malatya, Malmo-Lund, Manchester, Manila, Mansoura, Marburg, Medellin, Melbourne, Mexico City, Milan, Modena, Moscow, Muenster, Munich, Nancy, Nantes, New Delhi, New York, Newcastle (AUS), Newcastle u Tyne, Nijmegen, Nottingham, Omaha, Orlando, Oviedo, Oxford, Padua, Palermo, Panama, Paris, Pato Branco, Pecs, Perth, Pisa, Poitiers, Portland, Porto Alegre, Quebec, Regensburg, Ribeirao Preto, Rio de Janeiro, Rome, Rosario, Rostock, Rotterdam, Santa Fe, Santiago, Sao Paulo, Sheffield, Stuttgart, Sydney, Szeged, Tehran, Toronto, Tuebingen, Turin, Uppsala, Utrecht, Valencia, Vienna, Vilnius, Wellington, Winnipeg, Zagreb, Zurich.