Advertisement
The Journal of Heart and Lung Transplantation
International Society for Heart and Lung Transplantation.

Influence of human leukocyte antigen mismatching on bronchiolitis obliterans syndrome in lung transplantation

Published:September 22, 2015DOI:https://doi.org/10.1016/j.healun.2015.08.022

      Background

      Varying results have been reported in the investigation of human leukocyte antigen (HLA) mismatching and bronchiolitis obliterans syndrome (BOS) after lung transplantation (LTx).

      Methods

      The UNOS database was queried for the period 1997 to 2013 to examine HLA mismatching and its association with BOS in LTx.

      Results

      Of 16,959 first-time adult LTx recipients, 16,854 were included in the univariate Cox analysis and Kaplan–Meier survival function evaluation, and 14,578 were included in multivariate Cox models. Multivariate Cox analysis showed that the number of total HLA mismatches was significantly associated with greater hazard of BOS (HR = 1.060; 95% CI 1.013 to 1.108; p = 0.011), as was the presence of 2 HLA-A mismatches, when compared with 0 or 1 mismatch at that locus (HR = 1.128; 95% CI 1.026 to 1.240; p = 0.012). These results were confirmed using competing-risks regression models that adjusted for death before BOS diagnosis. Multivariate Cox models identified no significant association with BOS hazard for HLA-B (HR = 1.014; 95% CI 0.914 to 1.126; p = 0.785) or HLA-DR (HR = 1.085; 95% CI 0.987 to 1.193; p = 0.090) mismatches. Higher body mass index was associated with increased risk for BOS, whereas older age was protective against BOS. Induction with alemtuzumab (HR = 0.343; 95% CI 0.252 to 0.467; p < 0.001) or basiliximab (HR = 0.862; 95% CI 0.758 to 0.980; p = 0.023) and longer ischemic time (HR = 0.909; 95% CI 0.877 to 0.942; p < 0.001) were associated with lower hazard of BOS.

      Conclusions

      Total HLA mismatches are associated with increased risk for BOS, specifically at the A locus. Induction with alemtuzumab or basiliximab reduced the risk, whereas greater ischemic time appears to also be protective.

      Keywords

      Chronic lung allograft dysfunction, particularly bronchiolitis obliterans syndrome (BOS), which is the clinical counterpart of obliterative bronchiolitis (OB), has emerged as the major obstacle to long-term survival after lung transplantation (LTx).
      • Yusen R.D.
      • Edwards L.B.
      • Kucheryavaya A.Y.
      • et al.
      The registry of the International Society for Heart and Lung Transplantation: thirty-first adult lung and heart–lung transplant report—2014; focus theme: retransplantation.
      Within 5 years of LTx, 40.3% of first-time adult recipients develop BOS.
      • Yusen R.D.
      • Edwards L.B.
      • Kucheryavaya A.Y.
      • et al.
      The registry of the International Society for Heart and Lung Transplantation: thirty-first adult lung and heart–lung transplant report—2014; focus theme: retransplantation.
      Although the pathogenesis of BOS has not been completely characterized, multiple immunologic factors have been described as contributing to its development, including innate immunity response, cellular/humoral alloimmunity and cellular/humoral autoimmunity.
      • Gracon A.S.
      • Wilkes D.S.
      Lung transplantation: chronic allograft dysfunction and establishing immune tolerance.
      T-cell receptors recognize the allogeneic major histocompatibility complex (MHC), also termed human leukocyte antigen (HLA), on the donor cells via a direct pathway and peptide fragments of allogeneic MHC presented by recipient MHC molecules through an indirect pathway.
      • Gelman A.E.
      • Li W.
      • Richardson S.B.
      • et al.
      Cutting edge: acute lung allograft rejection is independent of secondary lymphoid organs.
      • Larsen C.P.
      • Morris P.J.
      • Austyn J.M.
      Donor dendritic leukocytes migrate from cardiac allografts into recipients’ spleens.
      Although the role of humoral alloimmunity in BOS is not as well recognized, anti-HLA antibodies after LTx are associated with increased risk of BOS and decreased survival.
      • Smith M.A.
      • Sundaresan S.
      • Mohanakumar T.
      • et al.
      Effect of development of antibodies to HLA and cytomegalovirus mismatch on lung transplantation survival and development of bronchiolitis obliterans syndrome.
      • Brugière O.
      • Thabut G.
      • Suberbielle C.
      • et al.
      Relative impact of human leukocyte antigen mismatching and graft ischemic time after lung transplantation.
      • Wisser W.
      • Wekerle T.
      • Zlabinger G.
      • et al.
      Influence of human leukocyte antigen matching on long-term outcome after lung transplantation.
      • Sundaresan S.
      • Mohanakumar T.
      • Smith M.A.
      • et al.
      HLA-A locus mismatches and development of antibodies to HLA after lung transplantation correlate with the development of bronchiolitis obliterans syndrome.
      • Schulman L.L.
      • Weinberg A.D.
      • McGregor C.
      • et al.
      Mismatches at the HLA-DR and HLA-B loci are risk factors for acute rejection after lung transplantation.
      • van den Berg J.W.
      • Hepkema B.G.
      • Geertsma A.
      • et al.
      Long-term outcome of lung transplantation is predicted by the number of HLA-DR mismatches.
      • Love R.B.
      • Meyer K.C.
      • Devito-Haynes L.D.
      • et al.
      Effect of HLA-DR mismatch on lung transplant outcome.
      • Chalermskulrat W.
      • Neuringer I.P.
      • Schmitz J.L.
      • et al.
      Human leukocyte antigen mismatches predispose to the severity of bronchiolitis obliterans syndrome after lung transplantation.
      • Quantz M.A.
      • Bennett L.E.
      • Meyer D.M.
      • et al.
      Does human leukocyte antigen matching influence the outcome of lung transplantation? An analysis of 3,549 lung transplantations.
      • Opelz G.
      • Süsal C.
      • Ruhenstroth A.
      • et al.
      Impact of HLA compatibility on lung transplant survival and evidence for an HLA restriction phenomenon: a collaborative transplant study report.
      Several studies have demonstrated HLA mismatching between donor and recipient enhancing the risk for BOS and reducing graft survival, but there is a lack of consistency in the reported involvement of specific loci.
      • Smith M.A.
      • Sundaresan S.
      • Mohanakumar T.
      • et al.
      Effect of development of antibodies to HLA and cytomegalovirus mismatch on lung transplantation survival and development of bronchiolitis obliterans syndrome.
      • Brugière O.
      • Thabut G.
      • Suberbielle C.
      • et al.
      Relative impact of human leukocyte antigen mismatching and graft ischemic time after lung transplantation.
      • Wisser W.
      • Wekerle T.
      • Zlabinger G.
      • et al.
      Influence of human leukocyte antigen matching on long-term outcome after lung transplantation.
      • Sundaresan S.
      • Mohanakumar T.
      • Smith M.A.
      • et al.
      HLA-A locus mismatches and development of antibodies to HLA after lung transplantation correlate with the development of bronchiolitis obliterans syndrome.
      • Schulman L.L.
      • Weinberg A.D.
      • McGregor C.
      • et al.
      Mismatches at the HLA-DR and HLA-B loci are risk factors for acute rejection after lung transplantation.
      • van den Berg J.W.
      • Hepkema B.G.
      • Geertsma A.
      • et al.
      Long-term outcome of lung transplantation is predicted by the number of HLA-DR mismatches.
      • Love R.B.
      • Meyer K.C.
      • Devito-Haynes L.D.
      • et al.
      Effect of HLA-DR mismatch on lung transplant outcome.
      • Chalermskulrat W.
      • Neuringer I.P.
      • Schmitz J.L.
      • et al.
      Human leukocyte antigen mismatches predispose to the severity of bronchiolitis obliterans syndrome after lung transplantation.
      • Quantz M.A.
      • Bennett L.E.
      • Meyer D.M.
      • et al.
      Does human leukocyte antigen matching influence the outcome of lung transplantation? An analysis of 3,549 lung transplantations.
      • Opelz G.
      • Süsal C.
      • Ruhenstroth A.
      • et al.
      Impact of HLA compatibility on lung transplant survival and evidence for an HLA restriction phenomenon: a collaborative transplant study report.
      With uncertain findings regarding HLA mismatching as a precursor for the development of BOS after LTx, we sought to assess the effect of total HLA mismatching as well as mismatches specifically at the A, B and DR loci using a database available in the United States. We hypothesized that HLA mismatching contributes to the development of BOS. With no previous research investigating the effect of induction immunosuppression on these HLA mismatching effects, we included contemporary induction agents in the adjusted models to determine their impact on the hazard risk (HR) of BOS associated with HLA mismatching, while also evaluating the role of graft ischemic time.

      Methods

       Data collection

      We retrospectively evaluated data from patients who were registered in the Organ Procurement and Transplant Network (OPTN) Standard Transplant Analysis and Research (STAR) thoracic database,

      United Network for Organ Sharing/Organ Procurement and Transplantation Network Standard Transplant Analysis and Research Database. http://optn.transplant.hrsa.gov/data/about/OPTNDatabase.asp. Accessed September 6, 2013.

      administered by United Network for Organ Sharing (UNOS). Our study was approved by institutional review board of The Ohio State University Wexner Medical Center with a waiver of the need for individual consent (IRB # 2012H0306). The UNOS/OPTN thoracic database was queried for all adult (≥18 years of age) patients between January 1997 and September 2013 who were first-time recipients of single or bilateral LTx from cadaveric donors. Patients with a history of re-transplantation (n = 33), patients induced with muromonab-CD3 (n = 170) and patients with incomplete data on HLA mismatch (n = 2,839) were excluded, leading to a sample size of 16,959. Patients who received muromonab-CD3 were excluded, as this agent is no longer available for clinical use.

       Statistical methods

      All analyses were performed using STATA/MP, version 13.1 (StataCorp LP, College Station, TX). Descriptive statistics for continuous variables are presented as mean and standard deviation, and descriptive statistics for categorical variables are presented as proportions. BOS onset was assessed at follow-ups scheduled for 6 months after LTx and each subsequent anniversary of LTx. BOS diagnosis was reported dichotomously (yes/no) until 2004, at which point differentiation between Grade 0p, Grade 1, Grade 2 and Grade 3 BOS was introduced. For consistency across years, BOS follow-up data were dichotomized as 1 = BOS diagnosed at follow-up (any grade, if applicable) vs 0 = BOS not yet diagnosed at any follow-up. Missing data on BOS onset were primarily due to attrition before the first follow-up. There were 2,458 patients (14.49%) with no BOS data who exited before the 6-month mark; 662 (3.90%) with no BOS data exiting after 6 months but before the next scheduled follow-up at the first anniversary of LTx; and 319 (1.88%) with no BOS data exiting 1 year or later after LTx. In the main analysis, cases of missing data on BOS were coded as not having been diagnosed with BOS at any follow-up. Re-analysis of the data set while excluding all 3,439 cases missing BOS data did not alter the findings reported (results available upon request).
      The main covariates of interest were total HLA mismatch level (continuous variable, with a range of 0 to 6) and mismatch at the A, B or DR locus, respectively (each dichotomized as 2 vs 0 or 1). Survival analysis was performed on time from LTx to the date of follow-up at which BOS was diagnosed, up to a maximum follow-up duration of 5 years (range of time to BOS diagnosis: 82 to 1,825 days). There were 105 cases with unknown or zero time spent at risk, and these were excluded from our analysis. Univariate survival analysis relating each covariate to the diagnosis of BOS after LTx was performed using Kaplan–Meier function curves and a log-rank test of differences in survival functions, and Cox proportional hazards models. Multivariate survival analysis was performed using Cox proportional hazards models to test the relationship between each main covariate and the hazard of developing BOS, controlling for recipient characteristics, donor characteristics and transplant characteristics. In Kaplan–Meier and Cox proportional hazards analyses of time to BOS onset, both death and loss to follow-up were treated as censoring events. To represent BOS onset and mortality before BOS onset as competing risks, multivariate competing-risks regression models were fitted using the Fine and Gray method.
      • Fine J.P.
      • Gray R.J.
      A proportional hazards model for the subdistribution of a competing risk.
      The exclusion of cases missing data on control variables reduced the sample size of the multivariate models to 14,578 cases.
      Cytomegalovirus (CMV) mismatching between recipient and donor may be associated with BOS and may coincide with HLA mismatching,
      • Smith M.A.
      • Sundaresan S.
      • Mohanakumar T.
      • et al.
      Effect of development of antibodies to HLA and cytomegalovirus mismatch on lung transplantation survival and development of bronchiolitis obliterans syndrome.
      but data on CMV mismatching are substantially incomplete in the UNOS database. Therefore, we performed a supplementary multivariate Cox proportional hazards analysis in a sub-sample of 8,863 patients, adding CMV mismatching as a covariate, where donor-negative (D) recipient-negative (R) was the reference group, and comparison groups included DR+, D+D and D+R+.

      Results

       Study population

      Table 1 summarizes patients’ demographics and characteristics of the sample used for analysis. Of 16,959 first-time, adult LTx recipients, only 18 (0.11%) had 0 mismatch whereas 637 (3.76%) had 2 or fewer mismatches. Due to the rarity of patients with no HLA mismatches, 0 and 1 mismatches were compared with 2 mismatches at each specific locus. There were 8,529 patients (50.29%) with 2 mismatches at the A locus, 11,926 patients (70.32%) with 2 mismatches at the B locus and 9,167 patients (54.05%) with 2 mismatches at the DR locus.
      Table 1Patients’ Demographics and Characteristics (N = 16,959)
      VariableN (%)Mean (SD)
      Total HLA mismatches4.60 (1.08)
       018 (0.11%)
       189 (0.52%)
       2530 (3.13%)
       31,971 (11.62%)
       44,489 (26.47%)
       56,193 (36.52%)
       63,669 (21.63%)
      Male recipient9,571 (56.44%)
      Male donor10,168 (59.96%)
      Race of recipient
       White14,582 (85.98%)
       Black1,287 (7.59%)
       Other1,090 (6.43%)
      Race of donor
       White11,222 (66.17%)
       Black2,863 (16.88%)
       Other2,874 (16.95%)
      Diagnosis
       PPH420 (2.48%)
       CF2,279 (13.44%)
       IPF5,747 (33.89%)
       COPD5,501 (32.44%)
       Sarcoidosis494 (2.91%)
       A1AD773 (4.56%)
       Other1,745 (10.29%)
      Induction medication
       None5,505 (32.46%)
       Basiliximab4,130 (24.35%)
       ALG, ATG1,564 (9.22%)
       Alemtuzumab978 (5.77%)
       Corticosteroids only4,418 (26.05%)
       Other364 (2.15%)
      Age (years)53.20 (12.94)
      Transplant year2006.27 (4.65)
      Creatinine (mg/dl)0.88 (0.52)
      BMI (kg/m2)24.65 (4.72)
      Ischemic time (hours)4.98 (1.73)
      Distance to center (miles/100)1.99 (2.39)
      FEV1 (% predicted)36.86 (21.11)
      FVC (% predicted)49.87 (17.95)
      6MWD (feet)832.54 (470.25)
      A1AD, α1-anti-trypsin deficiency, ALG, anti-lymphocyte globulin, ATG, anti-thymocyte globulin; BMI, body mass index; HLA, human leukocyte antigen; CF, cystic fibrosis; COPD, chronic obstructive pulmonary disease; FEV1, forced expiratory volume in 1 second; FVC, forced vital capacity; IPF, idiopathic pulmonary fibrosis; PPH, primary pulmonary hypertension; 6MWD, 6-minute walk distance.

       Univariate and Kaplan–Meier survival analysis

      Univariate Cox proportional hazards analysis showed that a 1-unit increase in total HLA mismatch level was significantly associated with a 7.7% greater hazard of developing BOS (HR = 1.077; 95% confidence interval [CI] 1.034 to 1.122; p < 0.001) (Table 2). As a sensitivity analysis, total HLA mismatch was dichotomized as 6 compared with ≤5, and the presence of 6 HLA mismatches was also associated with increased hazard of BOS (HR = 1.197; 95% CI 1.082 to 1.325; p = 0.001). Further analysis using univariate Cox models identified a significantly higher BOS hazard for HLA-A (HR = 1.162; 95% CI 1.065 to 1.268; p = 0.001), but not for HLA-B (HR = 1.074; 95% CI 0.975 to 1.182; p = 0.146) or HLA-DR (HR = 1.072; 95% CI 0.983 to 1.170; p = 0.117) (Table 2). Analysis of Kaplan–Meier survival functions showed significant differences in BOS-free survival by the total number of HLA mismatches (Figure 1) and by HLA-A mismatch (Figure 2, Figure 3), but not by HLA-B or HLA-DR mismatch (graphs not shown).
      Table 2Univariate Cox Survival Analysis of HLA Mismatches (N = 16,854)
      VariableNHR95% CIp-value
      Total HLA mismatch
      Continuous number of mismatches, range 0 to 6.
      16,8541.077(1.034 to 1.122)<0.001
      A locus mismatch 2
      Categorical variable with 0 or 1 mismatch as the reference group.
      16,8541.162(1.065 to 1.268)0.001
      B locus mismatch 2
      Categorical variable with 0 or 1 mismatch as the reference group.
      16,8541.074(0.975 to 1.182)0.146
      DR locus mismatch 2
      Categorical variable with 0 or 1 mismatch as the reference group.
      16,8541.072(0.983 to 1.170)0.117
      Male recipient16,8541.037(0.950 to 1.131)0.418
      Male donor16,8541.130(1.033 to 1.236)0.008
      Race of recipient16,854
       WhiteRef.
       Black1.134(0.966 to 1.331)0.123
       Other1.128(0.944 to 1.349)0.186
      Race of donor16,854
       WhiteRef.
       Black1.134(1.006 to 1.279)0.039
       Other1.086(0.964 to 1.224)0.175
      Diagnosis16,854
       PPHRef.
       CF1.215(0.879 to 1.679)0.239
       IPF1.411(1.032 to 1.927)0.031
       COPD1.175(0.861 to 1.605)0.309
       Sarcoidosis1.341(0.903 to 1.992)0.146
       A1AD1.149(0.801 to 1.649)0.450
       Other1.264(0.907 to 1.762)0.166
      Induction medication16,854
       NoneRef.
       Basiliximab0.978(0.870 to 1.100)0.715
       ALG, ATG0.917(0.787 to 1.069)0.267
       Alemtuzumab0.432(0.325 to 0.574)<0.001
       Corticosteroids only0.823(0.736 to 0.920)0.001
       Other0.964(0.741 to 1.255)0.786
      Age (years)16,8541.000(0.997 to 1.003)0.989
      Transplant year16,8541.096(1.083 to 1.109)<0.001
      Creatinine (mg/dl)16,6780.927(0.826 to 1.040)0.195
      BMI (kg/m2)16,8501.021(1.012 to 1.031)<0.001
      Ischemic time (hours)
      Continuous ischemic time centered at 5 hours.
      15,513
       Linear term0.957(0.929 to 0.985)0.003
       Quadratic term0.987(0.976 to 0.998)0.025
      Distance to center (miles/100)16,5951.013(0.994 to 1.032)0.167
      FEV1 (% predicted)16,0781.000(0.998 to 1.002)0.749
      FVC (% predicted)16,1060.997(0.995 to 0.999)0.025
      6MWD (feet)1,8440.9999(0.9997 to 1.0002)0.528
      Refer to Table 1 for abbreviations.
      a Continuous number of mismatches, range 0 to 6.
      b Categorical variable with 0 or 1 mismatch as the reference group.
      c Continuous ischemic time centered at 5 hours.
      Figure thumbnail gr1
      Figure 1Kaplan–Meier survival functions of time to bronchiolitis obliterans syndrome (BOS) by total HLA mismatch level (range 0 to 6) (N = 16,854). Log-rank test: chi-square (df = 6) = 25.86, p < 0.001.
      Figure thumbnail gr2
      Figure 2Kaplan–Meier survival functions of time to bronchiolitis obliterans syndrome (BOS) by A locus mismatch (range 0 to 2) (N = 16,854). Log-rank test: chi-square (df = 2) = 11.98, p = 0.003.
      Figure thumbnail gr3
      Figure 3Kaplan–Meier survival functions of time to bronchiolitis obliterans syndrome (BOS) by dichotomized A locus mismatch 2 vs 0 or 1 (N = 16,854). Log-rank test: chi-square (df = 1) = 11.46, p = 0.001.
      Several covariates were found to be significant in the univariate Cox models. Male donor gender and black donor race were associated with significantly higher hazard of BOS. The only diagnosis that increased the risk for BOS was idiopathic pulmonary fibrosis (IPF). An analysis of a contemporary list of induction agents demonstrated that induction therapy with either alemtuzumab or corticosteroids only (and no other induction therapy) reduced risk for BOS. Common practice in the United States is to give either an induction agent along with high-dose corticosteroids, or high-dose corticosteroids only, at time of LTx.
      • Yusen R.D.
      • Edwards L.B.
      • Kucheryavaya A.Y.
      • et al.
      The registry of the International Society for Heart and Lung Transplantation: thirty-first adult lung and heart–lung transplant report—2014; focus theme: retransplantation.
      Higher BMI and more recent LTx were associated with higher risk of BOS.
      In Cox proportional hazards analysis, the natural logarithm of the hazard rate is modeled as a linear function of covariates. The inclusion of both linear and quadratic terms for ischemic time tested whether ischemic time was non-linearly associated with the log hazard rate of BOS. The statistically significant HR of the quadratic term (HR = 0.987; 95% CI 0.995 to 0.999; p = 0.025) indicates that the log hazard rate of BOS was in fact non-linearly related to ischemic time. At the mean value of 5 hours, greater ischemic time was associated with lower hazard of BOS (HR = 0.957; 95% CI 0.929 to 0.985; p = 0.003). Yet, the quadratic function estimated for the relationship between ischemic time and the log hazard rate of BOS implies that, at ischemic durations between 1 and 3 hours, greater ischemic time was associated with greater hazard of BOS.

       Multivariate survival analysis

      The number of HLA mismatches continued to be associated with increased hazard of BOS in multivariate Cox proportional hazards analysis, adjusting for all covariates except 6-minute walk distance (HR = 1.060; 95% CI 1.013 to 1.108; p = 0.011) (Table 3, Model 1). Six-minute walk distance was not included in multivariate models due to very few patients having data on this variable (Table 2). Consistent with the result for the continuous number of total HLA mismatches, a multivariate Cox proportional hazards model showed that 6 total HLA mismatches continued to be significantly associated with increased hazard of BOS, relative to ≤5 total HLA mismatches (HR = 1.163; 95% CI 1.042 to 1.298; p = 0.007). In addition, multivariate Cox proportional hazards models regressing BOS on mismatches at specific loci showed that BOS hazard was significantly higher for HLA-A (HR = 1.128; 95% CI 1.026 to 1.240; p = 0.012) (Table 3, Model 2), but not for HLA-B (HR = 1.014; 95% CI 0.914 to 1.126; p = 0.785) or HLA-DR (HR = 1.085; 95% CI 0.987 to 1.193; p = 0.090) (data not shown).
      Table 3Multivariate Cox Proportional Hazards Analysis of Total HLA Mismatches and HLA Mismatches at the A Locus (N = 14,578)
      Model 1Model 2
      VariableHR95% CIp-valueHR95% CIp-value
      Total HLA mismatch
      Continuous number of mismatches, range 0 to 6.
      1.060(1.013 to 1.108)0.011
      A locus mismatch = 2
      Categorical variable with 0 or 1 mismatch as the reference group.
      1.128(1.026 to 1.240)0.012
      Male recipient0.960(0.861 to 1.070)0.4650.961(0.862 to 1.071)0.468
      Male donor1.141(1.025 to 1.270)0.0161.141(1.025 to 1.270)0.016
      Race of recipient
       WhiteRef.Ref.
       Black1.033(0.857 to 1.246)0.7351.036(0.860 to 1.250)0.708
       Other0.997(0.822 to 1.210)0.9750.995(0.820 to 1.208)0.960
      Race of donor
       WhiteRef.Ref.
       Black1.024(0.900 to 1.166)0.7151.028(0.904 to 1.171)0.671
       Other0.963(0.846 to 1.096)0.5710.969(0.851 to 1.102)0.631
      Diagnosis
       PPHRef.Ref.
       CF1.151(0.770 to 1.721)0.4921.149(0.769 to 1.718)0.498
       IPF1.245(0.847 to 1.830)0.2661.252(0.852 to 1.841)0.253
       COPD1.184(0.791 to 1.773)0.4111.184(0.791 to 1.772)0.413
       Sarcoidosis1.133(0.704 to 1.824)0.6061.135(0.706 to 1.827)0.601
       A1AD1.231(0.787 to 1.926)0.3621.237(0.790 to 1.931)0.355
       Other1.048(0.705 to 1.560)0.8161.049(0.705 to 1.560)0.814
      Induction medication
       NoneRef.Ref.
       Basiliximab0.862(0.758 to 0.980)0.0230.863(0.759 to 0.981)0.024
       ALG, ATG0.937(0.795 to 1.105)0.4410.940(0.797 to 1.108)0.458
       Alemtuzumab0.343(0.252 to 0.467)<0.0010.343(0.252 to 0.468)<0.001
       Corticosteroids only0.897(0.792 to 1.015)0.0850.898(0.793 to 1.016)0.089
       Other1.137(0.867 to 1.491)0.3521.138(0.868 to 1.492)0.349
      Age (years)0.990(0.984 to 0.995)<0.0010.990(0.984 to 0.995)<0.001
      Transplant year1.113(1.096 to 1.129)<0.0011.113(1.096 to 1.129)<0.001
      Creatinine (mg/dl)0.978(0.872 to 1.098)0.7080.978(0.871 to 1.097)0.699
      BMI (kg/m2)1.022(1.010 to 1.035)<0.0011.022(1.010 to 1.035)<0.001
      Ischemic time (hours)
      Continuous ischemic time, centered at 5 hours.
       Linear term0.909(0.877 to 0.942)<0.0010.909(0.877 to 0.942)<0.001
       Quadratic term0.993(0.981 to 1.005)0.2490.993(0.981 to 1.005)0.255
      Distance to center (miles/100)1.040(1.017 to 1.064)0.0011.040(1.017 to 1.064)0.001
      FEV1 (% predicted)0.997(0.993 to 1.001)0.1630.997(0.993 to 1.001)0.151
      FVC (% predicted)1.001(0.997 to 1.004)0.7931.001(0.997 to 1.004)0.769
      Refer to Table 1 for abbreviations.
      a Continuous number of mismatches, range 0 to 6.
      b Categorical variable with 0 or 1 mismatch as the reference group.
      c Continuous ischemic time, centered at 5 hours.
      In the multivariate Cox proportional hazards models, observations were censored at the date of death or loss to follow-up. Yet, death prior to BOS onset may be a competing risk to BOS diagnosis. Figure 4 shows Kaplan–Meier curves for death after LTx, stratified by the total number of HLA mismatches, with observations censored at BOS diagnosis or loss to follow-up. Log-rank test identified no significant differences in death before BOS diagnosis across HLA mismatch groups (p = 0.149). Table 4 shows a competing-risks regression model fitted using the same covariates as the multivariate Cox proportional hazards model. The left-hand panel shows sub-hazard ratios (SHR) for BOS onset, accounting for the competing risk of death before BOS onset, whereas the right-hand panel shows sub-hazard ratios for death before BOS onset, accounting for BOS onset as a competing risk. Competing-risks regression analysis confirmed a significant association between a higher number of HLA mismatches and increased risk of BOS after LTx (SHR = 1.050; 95% CI 1.003 to 1.099; p = 0.036). Competing-risks regression analysis also confirmed a significant association between mismatch at the HLA-A locus and increased risk of BOS (SHR = 1.114; 95% CI 1.014 to 1.224; p = 0.025) (Table 5).
      Figure thumbnail gr4
      Figure 4.Kaplan–Meier survival functions of time to death before bronchiolitis obliterans syndrome (BOS) diagnosis by total HLA mismatch level (range 0 to 6) (N = 16,854), log-rank test: chi-square (df = 6) = 9.47, p = 0.149.
      Table 4Competing-risks Regression Analysis of Total HLA Mismatches (N = 14,578)
      BOS onsetDeath before BOS onset
      VariableSHR95% CIp-valueSHR95% CIp-value
      Total HLA mismatch
      Continuous number of mismatches, range 0 to 6.
      1.050(1.003 to 1.099)0.0361.028(0.999 to 1.058)0.059
      Male recipient0.948(0.851 to 1.056)0.3331.059(0.986 to 1.137)0.115
      Male donor1.171(1.052 to 1.303)0.0040.903(0.843 to 0.967)0.003
      Race of recipient
       WhiteRef.Ref.
       Black1.048(0.869 to 1.263)0.6221.011(0.890 to 1.149)0.863
       Other1.015(0.837 to 1.232)0.8800.969(0.850 to 1.105)0.641
      Race of donor
       WhiteRef.Ref.
       Black0.950(0.834 to 1.081)0.4341.234(1.137 to 1.340)<0.001
       Other0.944(0.829 to 1.075)0.3831.087(0.998 to 1.183)0.057
      Diagnosis
       PPHRef.Ref.
       CF1.214(0.813 to 1.812)0.3440.920(0.714 to 1.185)0.518
       IPF1.385(0.943 to 2.036)0.0970.842(0.663 to 1.068)0.156
       COPD1.383(0.920 to 2.078)0.1190.718(0.561 to 0.919)0.008
       Sarcoidosis1.265(0.786 to 2.038)0.3330.828(0.609 to 1.125)0.227
       A1AD1.375(0.877 to 2.157)0.1650.772(0.584 to 1.021)0.070
       Other1.178(0.792 to 1.753)0.4190.842(0.658 to 1.078)0.172
      Induction medication
       NoneRef.Ref.
       Basiliximab0.896(0.788 to 1.018)0.0920.870(0.796 to 0.950)0.002
       ALG, ATG, thymoglobulin0.962(0.817 to 1.134)0.6450.947(0.848 to 1.058)0.337
       Alemtuzumab0.356(0.261 to 0.484)<0.0011.174(1.034 to 1.335)0.014
       Corticosteroids only0.895(0.791 to 1.012)0.0771.016(0.937 to 1.102)0.698
       Other1.246(0.950 to 1.635)0.1120.831(0.678 to 1.018)0.074
      Age (years)0.985(0.979 to 0.991)<0.0011.018(1.014 to 1.022)<0.001
      Transplant year1.081(1.067 to 1.096)<0.0010.970(0.962 to 0.978)<0.001
      Creatinine (mg/dl)0.918(0.807 to 1.044)0.1931.100(1.047 to 1.155)<0.001
      BMI (kg/m2)1.022(1.010 to 1.034)<0.0010.999(0.991 to 1.007)0.792
      Ischemic time (hours)
      Continuous ischemic time, centered at 5 hours.
       Linear term0.919(0.887 to 0.953)<0.0010.990(0.968 to 1.012)0.369
       Quadratic term0.988(0.975 to 1.000)0.0531.009(1.003 to 1.015)0.003
      Distance to center (miles/100)1.039(1.016 to 1.063)0.0010.993(0.979 to 1.008)0.364
      FEV1 (% predicted)0.997(0.993 to 1.001)0.1641.000(0.998 to 1.003)0.876
      FVC (% predicted)1.002(0.998 to 1.005)0.3960.997(0.995 to 1.000)0.041
      Refer to Table 1 for abbreviations.
      a Continuous number of mismatches, range 0 to 6.
      b Continuous ischemic time, centered at 5 hours.
      Table 5Competing-risks Regression Analysis of HLA Mismatches at the A Locus (N = 14,578)
      BOS onsetDeath before BOS onset
      VariableSHR95% CIp-valueSHR95% CIp-value
      A locus mismatch 2
      Categorical variable with 0 or 1 mismatch as the reference group.
      1.114(1.014 to 1.224)0.0251.048(0.985 to 1.114)0.136
      Male recipient0.949(0.852 to 1.057)0.3431.059(0.987 to 1.137)0.113
      Male donor1.171(1.052 to 1.304)0.0040.903(0.843 to 0.967)0.003
      Race of recipient
       WhiteRef.Ref.
       Black1.051(0.872 to 1.267)0.6021.014(0.893 to 1.152)0.831
       Other1.013(0.835 to 1.230)0.8920.969(0.850 to 1.105)0.635
      Race of donor
       WhiteRef.Ref.
       Black0.951(0.835 to 1.082)0.4431.239(1.142 to 1.345)<0.001
       Other0.948(0.833 to 1.079)0.4181.089(1.000 to 1.187)0.049
      Diagnosis
       PPHRef.Ref.
       CF1.213(0.812 to 1.811)0.3450.919(0.713 to 1.183)0.512
       IPF1.391(0.947 to 2.045)0.0930.844(0.665 to 1.071)0.162
       COPD1.384(0.921 to 2.080)0.1180.718(0.562 to 0.919)0.008
       Sarcoidosis1.269(0.787 to 2.044)0.3280.828(0.609 to 1.126)0.229
       A1AD1.381(0.881 to 2.166)0.1600.774(0.586 to 1.023)0.072
       Other1.179(0.792 to 1.755)0.4170.842(0.658 to 1.078)0.172
      Induction medication
       NoneRef.Ref.
       Basiliximab0.896(0.789 to 1.019)0.0950.870(0.797 to 0.951)0.002
       ALG, ATG0.964(0.818 to 1.135)0.6570.948(0.848 to 1.059)0.342
       Alemtuzumab0.356(0.261 to 0.484)<0.0011.175(1.034 to 1.335)0.013
       Corticosteroids only0.895(0.792 to 1.013)0.0791.017(0.938 to 1.102)0.688
       Other1.247(0.950 to 1.636)0.1110.832(0.679 to 1.019)0.076
      Age (years)0.985(0.979 to 0.990)<0.0011.018(1.014 to 1.022)<0.001
      Transplant year1.081(1.067 to 1.096)<0.0010.970(0.962 to 0.978)<0.001
      Creatinine (mg/dl)0.918(0.807 to 1.045)0.1961.099(1.047 to 1.154)<0.001
      BMI (kg/m2)1.022(1.010 to 1.034)<0.0010.999(0.991 to 1.007)0.793
      Ischemic time (hours)
      Continuous ischemic time, centered at 5 hours.
       Linear term0.919(0.887 to 0.953)<0.0010.990(0.968 to 1.012)0.370
       Quadratic term0.988(0.976 to 1.000)0.0551.009(1.003 to 1.015)0.003
      Distance to center (miles/100)1.039(1.016 to 1.063)0.0010.993(0.979 to 1.008)0.368
      FEV1 (% predicted)0.997(0.993 to 1.001)0.1621.000(0.998 to 1.003)0.885
      FVC (% predicted)1.002(0.998 to 1.005)0.3870.997(0.995 to 1.000)0.043
      Refer to Table 1 for abbreviations.
      a Categorical variable with 0 or 1 mismatch as the reference group.
      b Continuous ischemic time, centered at 5 hours.
      Several covariates were found to be consistent in the multivariate Cox models. As with the univariate Cox models, male donor gender was associated with a significantly higher hazard of BOS, although there were no longer significant differences in BOS hazard by donor race. Differences in BOS hazard across diagnostic categories were also not statistically significant in the multivariate models. Regarding induction agents, both basiliximab and alemtuzumab were found to be protective according to adjusted HR, whereas the use of corticosteroids only was no longer significantly related to BOS hazard. Older recipient age became associated with a decreased hazard of BOS, whereas this variable was not statistically significant in the univariate model. Higher BMI, more recent LTx, and longer distance from the donor hospital to the transplant center were associated with higher hazard of BOS. The quadratic transformation of ischemic time was no longer statistically significant, indicating that ischemic time was linearly related to the log hazard rate of BOS, with greater ischemic time predicting lower hazard of BOS when adjusted for other covariates.
      A supplementary multivariate Cox proportional hazards model was fitted adding CMV mismatch between donors and recipients as a covariate. In this multivariate model, a D+R+ CMV match was associated with increased BOS hazard, relative to DR (HR = 1.229; 95% CI 1.011 to 1.493; p = 0.038). The HR of the number of total HLA mismatches (HR = 1.075; 95% CI 1.012 to 1.142; p = 0.019) remained statistically significant after controlling for CMV mismatch, indicating that CMV mismatch was not a contributing factor in our analysis of HLA mismatching.

      Discussion

      The number of total HLA mismatches increased the risk for BOS, with mismatching at HLA-A specifically being associated with greater BOS hazard. Induction therapy with alemtuzumab and basiliximab at time of LTx was found to be protective using adjusted models. This protective effect of induction therapies at time of LTx merits their use in patients with HLA mismatches, with the need for further prospective research on their immunologic effects in relation to other factors at the time of transplant.
      The medical literature regarding HLA mismatching and its influence on BOS after LTx is not vigorous. The majority of the research consists of single-center studies or collaborations among a small number of transplant centers with a wide range of results regarding loci matching and development of BOS after LTx.
      • Smith M.A.
      • Sundaresan S.
      • Mohanakumar T.
      • et al.
      Effect of development of antibodies to HLA and cytomegalovirus mismatch on lung transplantation survival and development of bronchiolitis obliterans syndrome.
      • Brugière O.
      • Thabut G.
      • Suberbielle C.
      • et al.
      Relative impact of human leukocyte antigen mismatching and graft ischemic time after lung transplantation.
      • Wisser W.
      • Wekerle T.
      • Zlabinger G.
      • et al.
      Influence of human leukocyte antigen matching on long-term outcome after lung transplantation.
      • Sundaresan S.
      • Mohanakumar T.
      • Smith M.A.
      • et al.
      HLA-A locus mismatches and development of antibodies to HLA after lung transplantation correlate with the development of bronchiolitis obliterans syndrome.
      • Schulman L.L.
      • Weinberg A.D.
      • McGregor C.
      • et al.
      Mismatches at the HLA-DR and HLA-B loci are risk factors for acute rejection after lung transplantation.
      • van den Berg J.W.
      • Hepkema B.G.
      • Geertsma A.
      • et al.
      Long-term outcome of lung transplantation is predicted by the number of HLA-DR mismatches.
      • Love R.B.
      • Meyer K.C.
      • Devito-Haynes L.D.
      • et al.
      Effect of HLA-DR mismatch on lung transplant outcome.
      • Chalermskulrat W.
      • Neuringer I.P.
      • Schmitz J.L.
      • et al.
      Human leukocyte antigen mismatches predispose to the severity of bronchiolitis obliterans syndrome after lung transplantation.
      A few studies have attempted to further investigate HLA mismatching in larger databases. A study published in 2000 using data from the UNOS and International Society for Heart and Lung Transplantation registries found no significant correlation of HLA mismatching and BOS.
      • Quantz M.A.
      • Bennett L.E.
      • Meyer D.M.
      • et al.
      Does human leukocyte antigen matching influence the outcome of lung transplantation? An analysis of 3,549 lung transplantations.
      A more recent study
      • Opelz G.
      • Süsal C.
      • Ruhenstroth A.
      • et al.
      Impact of HLA compatibility on lung transplant survival and evidence for an HLA restriction phenomenon: a collaborative transplant study report.
      from 2010, using the Collaborative Transplant Study database,

      Collaborative Transplant Study. University of Heidelberg. http://www.ctstransplant.org/. Accessed November 13, 2014.

      showed that HLA mismatching at the A, B and DR loci influenced survival up to 5 years after LTx, but the investigators did not address the effect of HLA mismatching on BOS.
      Our findings provide useful information for clinicians caring for LTx recipients. Due to the higher risk for BOS in patients with HLA-A mismatches, induction immunosuppression with either basiliximab or alemtuzumab should be considered with closer monitoring of this population as they are at highest risk. Due to the study design, we cannot determine the mechanisms involved regarding the reasons why HLA-A mismatches have a stronger association with BOS compared with mismatches at other loci. Nevertheless, we have identified a relevant HLA disparity that needs further investigation.
      In comparison to the renal transplantation literature, our findings are consistent with growing evidence of variable immunogenicity with HLA antigens. The substantial variability in the immunogenicity of different HLA antigens is influenced by the presence or absence of cross-reactive antigens in the phenotype of patients.
      • Lucas D.P.
      • Leffell M.S.
      • Zachary A.A.
      Differences in immunogenicity of HLA antigens and the impact of cross-reactivity on the humoral response.
      There is a correlation between the number of amino acid differences between donor and recipient and antibody production.
      • Dankers M.K.
      • Witvliet M.D.
      • Roelen D.L.
      • et al.
      The number of amino acid triplet differences between patient and donor is predictive for the antibody reactivity against mismatched human leukocyte antigens.
      • Duquesnoy R.J.
      A structurally based approach to determine HLA compatibility at the humoral immune level.
      Although the number of amino acid differences may demonstrate some correlation with the antibody response, it does not completely define the immunogenicity and humoral response to specific mismatched antigens.
      • Lucas D.P.
      • Leffell M.S.
      • Zachary A.A.
      Differences in immunogenicity of HLA antigens and the impact of cross-reactivity on the humoral response.
      In addition to the number of amino acid sequences, the physiochemical properties of amino acid polymorphisms, including hydrophobicity and electrostatic charge between HLA Class I specificities, can predict HLA Class I alloantigen immunogenicity.
      • Kosmoliaptsis V.
      • Dafforn T.R.
      • Chaudhry A.N.
      • et al.
      High-resolution, three-dimensional modeling of human leukocyte antigen class I structure and surface electrostatic potential reveals the molecular basis for alloantibody binding epitopes.
      • Kosmoliaptsis V.
      • Chaudhry A.N.
      • Sharples L.D.
      • et al.
      Predicting HLA class I alloantigen immunogenicity from the number and physiochemical properties of amino acid polymorphisms.
      With antibodies reacting with conformational epitopes, the immunogenicity of the epitope is affected by these amino acid physiochemical properties of the amino acids.
      • Kosmoliaptsis V.
      • Dafforn T.R.
      • Chaudhry A.N.
      • et al.
      High-resolution, three-dimensional modeling of human leukocyte antigen class I structure and surface electrostatic potential reveals the molecular basis for alloantibody binding epitopes.
      With limited research investigating this concept in LTx, we cannot determine similarities in patients after LTx in comparison to renal transplant recipients. In a limited study of cardiothoracic transplant recipients who received a sequential kidney transplant, repeat HLA-A, HLA-B or HLA-DR mismatch occurred in 16 of 53 (30%) cases, but these mismatches were not associated with an adverse effect on kidney transplant outcome at 1 and 5 years.
      • Caskey F.J.
      • Johnson R.J.
      • Fuggle S.V.
      • et al.
      Renal after cardiothoracic transplant: the effect of repeat mismatches on outcome.
      Longer graft ischemic time has been associated with significantly reduced survival after LTx in the setting of higher levels of HLA mismatching, specifically at the A locus.
      • Brugière O.
      • Thabut G.
      • Suberbielle C.
      • et al.
      Relative impact of human leukocyte antigen mismatching and graft ischemic time after lung transplantation.
      Using a multivariate analysis, Brugière et al
      • Brugière O.
      • Thabut G.
      • Suberbielle C.
      • et al.
      Relative impact of human leukocyte antigen mismatching and graft ischemic time after lung transplantation.
      showed that the benefit of reduced mismatching of 0 or 1 at the HLA-A locus was equivalent to a benefit of reducing graft ischemic time by 168 minutes. These investigators examined only survival and not BOS. However, the results of the current study are at odds with the established medical literature, showing that longer graft ischemic time was protective against BOS, even after accounting for the competing risk of mortality before BOS diagnosis. Based on this discrepancy, the relationship between ischemic time and BOS clearly needs further study with defined patient populations.
      The present study is limited by several factors, including lack of data on the different phenotypes of chronic lung allograft dysfunction (CLAD). There are emerging data regarding CLAD phenotypes, including classical or fibrotic BOS, lymphocytic bronchiolitis, azithromycin-reversible allograft dysfunction and restrictive allograft syndrome,
      • Suwara M.I.
      • Vanaudenaerde B.M.
      • Verleden S.E.
      • et al.
      Mechanistic differences between phenotypes of chronic lung allograft dysfunction after lung transplantation.
      so our analysis could not differentiate these various categories. In addition, the data used were retrospectively collected and could have potentially excluded confounding variables. There was also a potential risk for data entry errors and missing data points that could have restricted statistical analysis. In particular, the timing of follow-up after LTx, with the first follow-up scheduled for 6 months post-transplant, meant that some patients exited the sample within 6 months before data on BOS onset could be collected. Furthermore, our study is limited due to inclusion of mismatches at the DQ and DP loci and not controlling for pre-existing anti-HLA donor-specific or non-specific antibodies in the recipients that could have altered the HR for BOS. Despite its limitations, this study draws results from a large, multi-institutional registry database of transplant recipients and thus reduces potential biases observed in single-institution observational studies.
      In conclusion, we found that HLA mismatches, specifically at the A locus, significantly increased the risk of BOS in LTx recipients. Induction immunosuppression at the time of LTx with either basiliximab or alemtuzumab was beneficial and mitigated the hazard risk for BOS. In contrast to published literature, longer graft ischemic time was associated with lower hazard of developing BOS after LTx. Based on our findings, prospective research is needed to delineate the interplay among variables at the time of LTx that are responsible for the later development of BOS.

      Disclosure statement

      The authors have no conflicts of interest to disclose.
      We thank Dmitry Tumin for statistical expertise in the data analysis.

      References

        • Yusen R.D.
        • Edwards L.B.
        • Kucheryavaya A.Y.
        • et al.
        The registry of the International Society for Heart and Lung Transplantation: thirty-first adult lung and heart–lung transplant report—2014; focus theme: retransplantation.
        J Heart Lung Transplant. 2014; 33: 1009-1024
        • Gracon A.S.
        • Wilkes D.S.
        Lung transplantation: chronic allograft dysfunction and establishing immune tolerance.
        Hum Immunol. 2014; 75: 887-894
        • Gelman A.E.
        • Li W.
        • Richardson S.B.
        • et al.
        Cutting edge: acute lung allograft rejection is independent of secondary lymphoid organs.
        J Immunol. 2009; 182: 3969-3973
        • Larsen C.P.
        • Morris P.J.
        • Austyn J.M.
        Donor dendritic leukocytes migrate from cardiac allografts into recipients’ spleens.
        Transplant Proc. 1990; 22: 1943-1944
        • Smith M.A.
        • Sundaresan S.
        • Mohanakumar T.
        • et al.
        Effect of development of antibodies to HLA and cytomegalovirus mismatch on lung transplantation survival and development of bronchiolitis obliterans syndrome.
        J Thorac Cardiovasc Surg. 1998; 116: 812-820
        • Brugière O.
        • Thabut G.
        • Suberbielle C.
        • et al.
        Relative impact of human leukocyte antigen mismatching and graft ischemic time after lung transplantation.
        J Heart Lung Transplant. 2008; 27: 628-634
        • Wisser W.
        • Wekerle T.
        • Zlabinger G.
        • et al.
        Influence of human leukocyte antigen matching on long-term outcome after lung transplantation.
        J Heart Lung Transplant. 1996; 15: 1209-1216
        • Sundaresan S.
        • Mohanakumar T.
        • Smith M.A.
        • et al.
        HLA-A locus mismatches and development of antibodies to HLA after lung transplantation correlate with the development of bronchiolitis obliterans syndrome.
        Transplantation. 1998; 65: 648-653
        • Schulman L.L.
        • Weinberg A.D.
        • McGregor C.
        • et al.
        Mismatches at the HLA-DR and HLA-B loci are risk factors for acute rejection after lung transplantation.
        Am J Respir Crit Care Med. 1998; 157: 1833-1837
        • van den Berg J.W.
        • Hepkema B.G.
        • Geertsma A.
        • et al.
        Long-term outcome of lung transplantation is predicted by the number of HLA-DR mismatches.
        Transplantation. 2001; 71: 368-373
        • Love R.B.
        • Meyer K.C.
        • Devito-Haynes L.D.
        • et al.
        Effect of HLA-DR mismatch on lung transplant outcome.
        J Heart Lung Transplant. 2001; 20: 177
        • Chalermskulrat W.
        • Neuringer I.P.
        • Schmitz J.L.
        • et al.
        Human leukocyte antigen mismatches predispose to the severity of bronchiolitis obliterans syndrome after lung transplantation.
        Chest. 2003; 123: 1825-1831
        • Quantz M.A.
        • Bennett L.E.
        • Meyer D.M.
        • et al.
        Does human leukocyte antigen matching influence the outcome of lung transplantation? An analysis of 3,549 lung transplantations.
        J Heart Lung Transplant. 2000; 19: 473-479
        • Opelz G.
        • Süsal C.
        • Ruhenstroth A.
        • et al.
        Impact of HLA compatibility on lung transplant survival and evidence for an HLA restriction phenomenon: a collaborative transplant study report.
        Transplantation. 2010; 90: 912-917
      1. United Network for Organ Sharing/Organ Procurement and Transplantation Network Standard Transplant Analysis and Research Database. http://optn.transplant.hrsa.gov/data/about/OPTNDatabase.asp. Accessed September 6, 2013.

        • Fine J.P.
        • Gray R.J.
        A proportional hazards model for the subdistribution of a competing risk.
        J Am Stat Assoc. 1999; 94: 496-509
      2. Collaborative Transplant Study. University of Heidelberg. http://www.ctstransplant.org/. Accessed November 13, 2014.

        • Lucas D.P.
        • Leffell M.S.
        • Zachary A.A.
        Differences in immunogenicity of HLA antigens and the impact of cross-reactivity on the humoral response.
        Transplantation. 2015; 99: 77-85
        • Dankers M.K.
        • Witvliet M.D.
        • Roelen D.L.
        • et al.
        The number of amino acid triplet differences between patient and donor is predictive for the antibody reactivity against mismatched human leukocyte antigens.
        Transplantation. 2004; 77: 1236-1239
        • Duquesnoy R.J.
        A structurally based approach to determine HLA compatibility at the humoral immune level.
        Hum Immunol. 2006; 67: 847-862
        • Kosmoliaptsis V.
        • Dafforn T.R.
        • Chaudhry A.N.
        • et al.
        High-resolution, three-dimensional modeling of human leukocyte antigen class I structure and surface electrostatic potential reveals the molecular basis for alloantibody binding epitopes.
        Hum Immunol. 2011; 72: 1049-1059
        • Kosmoliaptsis V.
        • Chaudhry A.N.
        • Sharples L.D.
        • et al.
        Predicting HLA class I alloantigen immunogenicity from the number and physiochemical properties of amino acid polymorphisms.
        Transplantation. 2009; 88: 791-798
        • Caskey F.J.
        • Johnson R.J.
        • Fuggle S.V.
        • et al.
        Renal after cardiothoracic transplant: the effect of repeat mismatches on outcome.
        Transplantation. 2009; 87: 1727-1732
        • Suwara M.I.
        • Vanaudenaerde B.M.
        • Verleden S.E.
        • et al.
        Mechanistic differences between phenotypes of chronic lung allograft dysfunction after lung transplantation.
        Transpl Int. 2014; 27: 857-867