OUTCOMES The Influence of HLA Mismatches and Immunosuppression on Kidney Graft Survival: An Analysis of More Than 1300 Patients L. Martins, I. Fonseca, S. Sousa, C. Matos, J. Santos, L. Dias, A.C. Henriques, A.M. Sarmento, and A. Cabrita ABSTRACT New immunosuppressive drugs used in kidney transplantation decreased the incidence of acute rejection. It was hypothesized that, with their power, the importance of HLA matching was decreased. To evaluate the influence of HLA matching, immunosuppression, and other possible risk factors, we analyzed data of 1314 consecutive deceased donor kidney transplantation. We divided the patient population into 4 cohorts, according to the era of transplantation: era 1, before 1990, azathioprine (Aza) and cyclosporine (Csa) no microemulsion; era 2, between 1990 and 1995, Csa microemulsion; era 3, between 1996 and 2000, wide use of mycophenolate mofetil (MMF) and anti-thymocyte globulin (ATG); and era 4, after 2000, marked by sirolimus and tacrolimus (TAC) use. Multivariate analysis compared death-censored graft survival. Using as reference the results obtained with 0 HLA mismatches, we verified, during era 1 and era 2, an increased risk of graft loss for all of the subgroups with HLA mismatch ⬎0. However, during era 3 and era 4, the number of HLA mismatches did not influence graft survival. Although acute rejection and delayed graft function, which decreased in the later periods, remained as prognostic factors for graft loss. Considering the immunosuppressive protocol with Csa⫹Aza⫹Pred as reference, protocols used after 1995 with Pred⫹Csa⫹ATG, with Pred⫹Csa⫹MMF, and with Pred⫹Tac⫹MMF presented better survival results. Results showed that the significance of HLA matching decreased while the results improved with the new immunosuppressant drugs. These observations support the hypothesis that the weakened importance of HLA matching may be a consequence of the increasing efficacy of the immunosuppression. I T WAS shown several years ago, by the Collaborative Transplant Study, that there is a negative influence of human leukocyte antigen (HLA) mismatch between donor and recipient on kidney graft survival.1 In this study, survival curves were clearly separated, corresponding better survival curves to lower number of HLA mismatches. Meanwhile, much has changed, especially in terms of immunosuppression. It is possible that, with the more From the Nephrology Department, Renal Transplant Unit, Hospital Geral de Santo António, Oporto, Portugal. Address reprint requests to La Salete Martins, Nephrology Department, Hospital Santo António, Largo Professor Abel Salazar, 4050-011 Porto, Portugal. E-mail: lasalete@clix.pt © 2007 by Elsevier Inc. All rights reserved. 360 Park Avenue South, New York, NY 10010-1710 0041-1345/07/$–see front matter doi:10.1016/j.transproceed.2007.07.033 Transplantation Proceedings, 39, 2489 –2493 (2007) 2489 2490 MARTINS, FONSECA, SOUSA ET AL powerful immunosuppressive drugs, the impact of HLA mismatch on kidney graft survival has decreased. It may be not so important to get high-matched donor and recipient pairs. There are some recently published studies sustaining the negative impact of HLA mismatches on graft survival2 and others showing the lack of this impact.3 In the present study, we investigated the impact of HLA mismatches on kidney transplant (KT) allograft survival, considering the evolution of immunosuppressive protocols over the years. A multivariate analysis was then achieved, using a Cox proportional hazards model to identify factors that were associated with graft survival and considering those 4 KT groups. Covariates included patient demographics (age and gender), clinical factors (time on dialysis and previous transplantation), as well as donor factors (age, antigen matching, and gender), acute rejection and delayed graft function occurrence, and immunosuppressive protocol used. Hazard ratios (HR) are presented with 95% confidence intervals (CI). All analyses were performed using SPSS software version 14.0 (SPSS, Inc., Chicago, Ill, United States), and a P value ⬍ .05 was considered significant. PATIENTS AND METHODS RESULTS This retrospective analysis included all deceased donor KT performed in our center (n ⫽ 1314) from July 1983 to December 2005. Recipients of grafts from live donors (n ⫽ 47) and multiple-organ grafts (pancreas-kidney [n ⫽ 46] and liver-kidney [n ⫽ 6]) were excluded. Variables analyzed included patient age, gender, race, cause of chronic renal failure, time on dialysis, last panel-reactive antibody (PRA), donor age, number of HLA mismatches, cold ischemia time, immunosuppressive regimen used, incidence of acute rejection, and delayed graft function. Acute rejection was defined as treated acute rejection, with or without confirmatory histology. Delayed graft function was defined as the need for dialysis during the first week after transplantation, excluding other causes. Graft loss was defined as the need for permanent dialysis or repeated transplantation. To evaluate the influence of HLA mismatches on graft outcome a primary analysis was done in the overall cohort. Subsequently, the study population was divided into 4 groups, based on times in which immunosuppressive medications were introduced into clinical practice: era 1, before and during 1990, the era of Azathioprine (Aza) and cyclosporine (CsA), no microemulsion, n ⫽ 226 (17.2%); era 2, between 1991 and 1995, the era of CsA microemulsion, n ⫽ 420 (32.0%); era 3, between 1996 and 2000, marked by mycophenolate mofetil (MMF) introduction and by the wide use of anti-thymocyte globulin (ATG), n ⫽ 383 (29.1%); and era 4, after 2000, the time of sirolimus availability and wide use of tacrolimus (Tac), n ⫽ 285 (21.7%). The 1314 KT studied (59.4% men) had a mean age of 38 ⫾ 14 years, and almost all were Caucasians (99.4%). They had a mean time on dialysis of 47.4 ⫾ 40.6 months, it was the first transplantation for 88.8% of patients, and diabetes was the cause of renal failure in 51 KT patients (3.8%). Mean donor age was 31 ⫾ 16 years, and the number of HLA mismatches was 3 ⫾ 1. Sixty KT patients had 0 mismatches (MM) (4.9%), 92 had 1 MM (7.5%), 242 (19.6%) had 2 MM, 310 (25.1%) had 3 MM, 313 (25.4%) had 4 MM, 158 (12.8%) had 5 MM, and 58 (4.7%) had 6 MM. Mean last PRA of the recipient was 3.4 ⫾ 11.9% (16% of the patients had last PRA levels ⬎50%) and mean cold ischemia time was 23.4 ⫾ 6.9 hours. Delayed graft function occurred in 36.4% and acute rejection in 26.8%. Global patient, graft, and death-censored graft survival rates were 96.6%, 87.7%, and 90.5%, respectively, at 1 year; 92.8%, 79.4%, and 84.8%, respectively, at 5 years; 85.9%, 66.7%, and 76.2%, respectively, at 10 years; and 72.6%, 42.6%, and 60.2%, respectively, at 20 years. In the univariate survival analysis on the overall cohort, the number of HLA mismatch (0 to 6, as a continuous variable) had no significant effect on graft failure (P ⫽ .791). Considering HLA mismatch as a categorical variable (0 to 6, with 0 mismatches as the reference group), only transplantations with 6 and 1 mismatches were significantly associated with graft failure (HR ⫽ 2.89; 95% CI, 1.1–5.6; P ⫽ .016, for 6 mismatches and HR ⫽ 2.42; 95% CI, 1.2– 6.9; P ⫽ .039 for 1 mismatch). Subsequent analysis was done after stratifying the study population into 4 groups according to the era of transplantation. Patients belonging to era 1 and 2 had an increased risk of graft loss for almost all subgroups with HLA mismatch, comparing to 0 mismatches as the reference category (1 MM, HR ⫽ 4.09; 95% CI, 1.2–13.9; P ⫽ .024; 2 MM, HR ⫽ 4.43; 95% CI, 1.4 –14.2; P ⫽ .012; 3 MM, HR ⫽ 4.48; 95% CI, 1.4 –14.3; P ⫽ .012; and 6 MM, HR ⫽ 5.58; 95% CI, 1.6 –19.2; P ⫽ .006). This negative impact on graft outcome was statistically significant for almost all HLA mismatches categories (1, 2, 3, and 6), and nearly significant in the remaining categories (4 MM, HR ⫽ 3.23; 95% CI, 1.0 –10.4; P ⫽ .05; and 5 MM, HR ⫽ 3.22; 95% CI, 1.0 –10.6; P ⫽ .056) (Fig 1). For KT performed during era 3 and 4 (Fig 2) none of the HLA mismatch categories (0 to 6) significantly influenced graft survival, also considering 0 Statistical Analysis Baseline demographic and transplantation factors are described as means ⫾ SD for continuous variables and as frequency distributions for categorical variables. Statistical significance of the differences between groups was tested using two-sample t test for continuous variables and chi-square test for categorical variables. Patient survival, uncensored graft survival, and graft survival censored for patient death were determined using the KaplanMeier survival estimates, and differences between survival curves were assessed using the log-rank test. The influence of HLA mismatches on graft outcome was studied using univariate and multivariate Cox regression analysis, using graft survival censored for patients who died with a functioning graft. A primary univariate survival analysis was done in the overall cohort to evaluate the impact of HLA mismatches on graft survival. Given the possibility of changes in the relationship between HLA mismatches and graft survival, due to modifications in treatment patterns over the years, the cohort was stratified into 4 groups according to the era of transplantation and a secondary univariate survival analysis was done. KIDNEY GRAFT SURVIVAL 2491 Fig 1. Death-censored graft survival according to number of HLA mismatches (era 1 and 2, from 1983 to 1995). mismatches as the reference category (1 MM, P ⫽ .888; 2 MM, P ⫽ .441; 3 MM, P ⫽ .448; 4 MM, P ⫽ .851; 5 MM, P ⫽ .636; 6 MM, P ⫽ .671). However, in these 2 latter eras, the occurrence of delayed graft function and acute rejection remained as significant negative predictors for graft survival (HR ⫽ 3.15; 95% CI, 1.74 –5.71; P ⬍ .001; and HR ⫽ 2.18; 95% CI, 1.21–3.95; P ⫽ .01), as well as donor age (HR ⫽ 1.03; 95% CI, 1.01–1.05; P ⫽ .014), adjusted for recipient age, gender, time on dialysis, and existence or not of a previous transplantation. Another covariate included was a Fig 2. Death-censored graft survival according to number of HLA mismatches (era 3 and 4, from 1996 to 2005). KT performed from a female donor to a male recipient, but it had no influence on graft survival. Comparing era 1 and 2 to era 3 and 4, donor age was superior in the later periods (34 ⫾ 16 vs. 28 ⫾ 15 years; P ⬍ .001), as well as recipient age (39.9 ⫾ 13.9 vs 36.3 ⫾ 12.6 years; P ⬍ .001), but was lower than the incidence of delayed graft function (32.8% vs 39.9%; P ⫽ .008) and acute rejection (16.8% vs 37.5%; P ⬍ .001). Deathcensored graft survival rates were significantly better for era 3 and 4 than for era 1 and 2 (92%, 88%, and 84% for era 3 2492 MARTINS, FONSECA, SOUSA ET AL and 4 vs 88%, 80%, and 71% for era 1 and 2, respectively, for 1, 5, and 10 years; P ⬍ .001). No significant differences were found between era 1 and 2 and era 3 and 4 comparing the proportion of patients with 0 mismatches (5.1% vs 4.0%; P ⫽ .427), mean number of HLA mismatches (3.18 ⫾ 1.51 vs 3.14 ⫾ 1.38; P ⫽ .645), distribution of recipient gender (male 58.2 vs 60.8%; P ⫽ .371), and time on dialysis (3.9 ⫾ 4.0 vs 4.0 ⫾ 5.6 years; P ⫽ .717). Diabetes as cause of chronic renal failure was more prevalent in era 3 and 4 (5.4% vs 2.3%; P ⫽ .004). The frequency of recipients with last PRA ⬎50% was significantly higher on era 1 and 2 versus era 3 and 4 (26.0% vs 6.3%; P ⬍ .001), but the proportion of second transplantations was significantly higher in era 3 and 4 (7.1% vs 12.3%; P ⫽ .002), along with shorter cold ischemia time (24.5 ⫾ 5.0 vs 22.0 ⫾ 8.3 hours; P ⱕ .001). The effect of immunosuppressive protocols on graft survival was also assessed. Considering the immunosuppressive protocol with prednisone (Pred)⫹Csa⫹Aza as reference, the protocols associated with better graft survival were Pred⫹Csa⫹ATG (HR ⫽ 0.55; 95% CI, 0.35– 0.88; P ⫽ .013); Pred⫹Csa⫹MMF (HR ⫽ 0.52; 95% CI, 0.31– 0.89; P ⫽ .017); and Pred⫹Tac⫹MMF (HR ⫽ 0.28; 95% CI, 0.11– 0.75; P ⫽ .011). Protocol with Pred⫹Aza presented poor survival (Table 1). DISCUSSION As reported by Hariharan,4 we also experienced continuous improvement of our results on KT since the start on 1983. Unlikely data recently published by Meier-Kriesche,5 graft survival outcomes at our center after 1995, are better than those obtained during the previous period. This improvement was achieved even with older donors and recipients. In fact, as at other centers, we also extended criteria for inclusion of candidates on the transplantation waiting list, accepting older patients and those with more comorbid situations, such as diabetics, and also marginal donors. We observed a decreasing incidence of acute rejection and delayed graft function in the later period, both probably contributing to the better results obtained. During the period from 1983 to 1995, multivariate analysis showed an unequivocal benefit of HLA matching. This is clearly showed by the widening of the survival curves according to the number of HLA mismatches (Fig. 1). However, after 1995, the curves are much closer together and the survival rates according to the number of HLA mismatches are no longer significantly different, whatever number it was, compared with 0 mismatches (Fig 2). Su et al3 showed a progressive decline of HLA matching significance in KT and our results are in accordance with these data. Policy for organ allocation in Portugal includes national sharing for full-matched KT. United Network for Organ Sharing (UNOS) experience in the United States6 is in favor of national sharing because better outcomes were obtained from full-matched KT, with little increase in cold ischemia time. Longer cold ischemia time may overpass the benefit of full matching and national allocation.3,7 The protective effect of HLA matching is more controversial for partially matched KT and it must be weighed against the disadvantages of the extended cold ischemia time. With the results here presented we could not prove the advantage of HLA matching after 1995, although we have not observed an increase in cold ischemia time, on the contrary, it has been reduced. After 1995 we observed a lower frequency of patients with a PRA ⬎50%, but in terms of immunological risk, it may be balanced by the increment of second transplantations; the proportion of KT with 0 mismatches was similar in both periods. Several other factors may affect graft survival after KT. The higher power of the most recent immunosuppressive drugs (ATG, MMF, Tac, and Sirolimus) may be one of the reasons for the weakening of the impact of HLA matching. A previous report about MMF8 showed that its use in KT does not obviate the effects of HLA matching and vice versa; both contribute to the decrease of graft loss. Improving results on KT presented by Hariharan et al4 could not Table 1. The Effect of Immunosuppressive Protocols on Graft Survival Variables in the Equation 95.0% CI for Exp(B) Pred⫹Csa⫹Aza (reference protocol n ⫽ 52) Pred⫹CsA (n ⫽ 330) Pred⫹ATG (n ⫽ 21) Pred⫹CsA⫹ATG (n ⫽ 318) Pred⫹Aza (n ⫽ 33) Pred⫹CsA_ATG_Aza (n ⫽ 142) Pred⫹CsA⫹MMF (n ⫽ 198) Pred⫹CsA_MMF_anti_CD25 (n ⫽ 61) Pred⫹Tac_MMF_ATG (n ⫽ 24) Pred⫹Tac⫹MMF (n ⫽ 79) Pred⫹CsA_Rapa (n ⫽ 9) Others (n ⫽ 46) B SE Wald df Sig Exp(B) Lower Upper ⫺.044 .046 ⫺.597 .575 ⫺.327 ⫺.653 ⫺.186 ⫺.934 ⫺1.270 ⫺.595 ⫺1.420 .228 .370 .239 .290 .259 .273 .351 .740 .498 .739 .741 51.105 .037 .015 6.229 3.936 1.602 5.744 .280 1.592 6.513 .648 3.675 11 1 1 1 1 1 1 1 1 1 1 1 .000 .848 .901 .013 .047 .206 .017 .597 .207 .011 .421 .055 .957 1.047 .551 1.777 .721 .520 .830 .393 .281 .552 .242 .612 .507 .345 1.007 .434 .305 .417 .092 .106 .130 .057 1.496 2.163 .880 3.135 1.197 .888 1.653 1.676 .745 2.347 1.032 KIDNEY GRAFT SURVIVAL be explained by Tac or MMF, because of the small number of patients treated with these drugs in this analysis. The analysis of the impact of the global immunosuppressive protocol may be more elucidative than analyzing the impact of a single immunosuppressive agent. Our data suggest protocols comprising Pred⫹CsA⫹ATG or Pred⫹CsA⫹MMF or Pred⫹Tac⫹MMF have better results than the protocol considered as reference with Pred⫹CsA⫹Aza. The poor results with only Pred⫹Aza, used in the first years, were expected. Because the wide use of ATG, MMF, and Tac at our center occurred after 1995, it is possible that the beneficial effect of these immunosuppressive drugs could have attenuated or annulated the impact of HLA matching on graft survival, as observed in the same period. In conclusion, in our experience, since 1995 HLA matching in KT is no longer a risk factor for long-term graft survival, whereas acute rejection, delayed graft function, and donor age are still risk factors. Immunosuppressive protocols including more recent and powerful drugs, such as ATG, Tac, and MMF, correlated with better long-term survivals. This may support our hypothesis that the efficacy of the immunosuppression decreased the importance of HLA matching. 2493 REFERENCES 1. Opelz G for the Collaborative Transplant Study: Correlation of HLA matching with kidney graft survival in patients with or without cyclosporine treatment. Transplantation 40:240, 1985 2. Opelz G for the Collaborative Transplant Study: New immunosuppressants and HLA matching. Transplant Proc 33: 467, 2001 3. Su X, Zenios SA, Chakkera H, et al: Diminishing significance of HLA matching in kidney transplantation. Am J Transplant 4:1501, 2004 4. Hariharan S, Johnson CP, Bresnahan BA, et al: Improved graft survival after renal transplantation in the United States, 1988 to 1996. N Engl J Med 342:605, 2000 5. Meier-Kriesche HU, Schold JD, Srinivas TR, et al: Lack of improvement in renal allograft survival despite a marked decrease in acute rejection rates over the most recent era. Am J Transplant 4:378, 2004 6. Takemoto SK, Terasaki PI, Gjertson DW, et al: Twelve years’ experience with national sharing of HLA-matched cadaveric kidneys for transplantation. N Engl J Med 343:1078, 2000 7. Lee CM, Carter JT, Alfrey EJ, et al: Prolonged cold ischemia time obviates the benefits of 0 HLA mismatches in renal transplantation. Arch Surg 135:1016, 2000 8. Meier-Kriesche HU, Ojo AO, Leichtman, et al: Interaction of mycophenolate mofetil and HLA matching on renal allograft survival. Transplantation 71:398, 2001