Humoral alloreactivity is well established to predict adverse allograft outcomes. However, in some recipients, alloantibodies may also occur in the absence of graft dysfunction. We evaluated if and how often complement- and noncomplement-fixing alloantibodies are detectable in stable recipients and whether, in this context, they affect long-term outcomes. Sera obtained from 164 kidney transplant recipients at 2, 6 and 12 months were evaluated by FlowPRA screening and single-antigen testing for detection of IgG- or C4d-fixing HLA panel reactivity and donor-specific antibodies (DSA). Applying stringent criteria, we selected 34 patients with an uneventful 1-year course (no graft dysfunction or rejection) and excellent graft function at 12 months [estimated glomerular filtration rate (eGFR) ≥60 mL/min and proteinuria ≤0.5 g/24 h]. Nine (27%) and 5 (15%) of these recipients tested positive by [IgG] and [C4d]FlowPRA screening, respectively. In five cases, DSA were identified. Frequencies of positive test results and DSA binding intensities were not significantly lower than those documented for patients who did not fulfill the above criteria. In recipients with an excellent 1-year course, FlowPRA reactivity was not associated with lower eGFR or increased protein excretion during 68-month median follow-up. Our results suggest cautious interpretation of antibody monitoring in patients with normal-functioning grafts.
Numerous studies have shown that the presence of preexisting or newly formed circulating alloantibodies predicts adverse short- and long-term kidney allograft outcomes (1,2). A causal relationship between alloantibody formation and rejection was reinforced by studies demonstrating tight statistical associations between detectable alloantibodies and C4d deposition along transplant capillaries, a footprint of antibody-triggered classical complement activation (3–6). At the same time, however, there is increasing evidence that donor antigen-specific antibodies do not inevitably cause graft injury (7). Indeed, in the setting of ABO-incompatible transplantation, desensitized recipients were found to maintain stable transplant function despite recurrence of detectable donor blood group-specific antibodies (8–10). It was suggested that, in accordance with the results obtained in experimental xenotransplantation models (7), such antidonor reactivity could reflect transplant accommodation, a state of acquired resistance to immune-mediated injury (8–10). At a molecular level, various underlying mechanisms have been discussed, including altered complement regulation, disruption of cellular adhesion, or other mechanisms promoting cytoprotection (11–14). For ABO-compatible transplantation, there is less published evidence for antibody formation in the absence of graft dysfunction. Some highly sensitized recipients subjected to pretransplant apheresis were found to maintain stable graft function despite postoperative recurrence of donor-specific antibodies (DSA) (15,16). Moreover, in a large protocol biopsy study, Mengel et al. (17) found that capillary C4d staining may also occur in stable kidney allografts. Interestingly, at least in the short term, such C4d staining was not associated with adverse graft function and survival (17). In accordance with these results, which may point to a role of subclinical humoral alloimmunity, some authors have reported on circulating HLA alloantibodies in patients with stable kidney allograft function (18,19). The true incidence of clinically silent antibody formation and its actual long-term clinical impact, however, are not well known and remain to be established.
Applying modified flow cytometric solid-phase testing for identification of complement (C4d)- and/or noncomplement-fixing (IgG) anti-HLA alloantibodies (6,20), we addressed the question if and how often posttransplant alloantibodies can be detected in patients on standard immunosuppressive therapy showing a quite uneventful clinical course (no documented rejection or graft dysfunction) and excellent graft function at 1 year. In an attempt to quantify and characterize detected alloreactivities, a major objective of this study was to evaluate, whether, in this context, HLA reactivity predicts inferior long-term graft outcome.
Patients and Methods
In this retrospective analysis, an overall cohort of 164 recipients of a deceased donor kidney allograft (transplantation between 2000 and 2004), who had been alive with a functioning graft for at least 12 months, was subjected to serial serological evaluation. For included recipients, adequately stored posttransplant serum samples obtained at 2, 6 and 12 months as well as cryopreserved donor splenocytes were available. Baseline patient characteristics are detailed in Table 1. The majority of included patients received calcineurin inhibitor-based immunosuppression. Forty-six patients (28%) received antibody induction with anti-IL-2 receptor monoclonal antibody or a depleting antilymphocyte antibody. Ten recipients with broad complement-dependent cytotoxicity (CDC)-panel-reactive antibody (PRA) reactivity had been subjected to desensitization by peritransplant immunoadsorption according to a previously published protocol (21). Graft outcomes including levels of estimated glomerular filtration rate (eGFR) calculated by the Mayo Clinic equation (22) are shown in Table 2.
1Criteria for patient selection are detailed in the Results section.
2The overall cohort included patients who survived ≥1 year with a functioning graft and for whom adequate recipient and donor material for a comprehensive serological evaluation was available.
Follow-up (months), median (IQR)
eGFR (mL/min), median (IQR)
at 1 year
at 4 years
SrCr (mg/dL), median (IQR)
at 1 year
at 4 years
24-h proteinuria (g), median (IQR)
at 1 year
at 4 years
Acute cellular rejection during follow-up, N (%)
C4d+ graft dysfunction during follow-up, N (%)
4-year overall graft survival (%)
4-year patient survival (%)
Sera were evaluated for C4d-fixing and non-C4d-fixing HLA reactivities applying modified FlowPRA screening and single-antigen (SA) testing.
Using the FlowPRA screening test (One Lambda, Canoga Park, CA), IgG ([IgG]FlowPRA) and C4d-fixing panel reactivities ([C4d]FlowPRA) were simultaneously assessed by multicolor staining as previously described (6). In brief, FlowPRA beads were incubated with test sera for 30 min at 4°C, followed by 30 min incubation with serum obtained from a nonsensitized healthy male volunteer as human complement source. Beads were washed and stained with pretitered allophycocyanin (APC)-conjugated polyclonal donkey antihuman IgG antibody (Jackson Immuno Research Europe, Soham, UK) and fluorescein isothiocyanate (FITC)-conjugated rabbit polyclonal anti-C4d antibody (Biomedica, Vienna, Austria). Using a FACSCalibur flow cytometer (Becton Dickinson, San Jose, CA), HLA class I and II beads were separated by FL2 fluorescence, and IgG binding was analyzed on FL4 and C4d deposition on FL1 histograms. Markers were set according to a nonbinding control serum. Results are expressed as a percentage of positive events (% panel reactivity). A sample was classified FlowPRA positive if panel reactivity was ≥10%.
For characterization of HLA specificities, sera were evaluated by FlowPRA HLA class I or HLA class II single-antigen detection tests (One Lambda). IgG and C4d binding was detected by the protocol described above for FlowPRA screening. SA beads were distinguished according to their individual FL2 fluorescence. Markers were set according to binding of a negative control. A sample was classified positive if ≥20% of the events shifted to the right of the marker.
Selected serum samples were further subjected to flow cytometry crossmatch (FCXM) testing. In brief, cryopreserved spleen cells were incubated with test serum at room temperature for 30 min. The cells were washed and incubated with pretitered FITC-conjugated goat antihuman IgG (Accurate Chemical and Scientific Co., Westbury, NY), APC-labeled CD3 and peridin-chlorophyll-a-protein-labeled CD19 monoclonal antibodies (Becton Dickinson). IgG antibody binding to CD3+ T cells (T cell FCXM) or CD19+ B cells (B cell FCXM) was analyzed on FL2 histograms. The FCXM was considered positive when the fluorescence intensity was greater than the mean plus two standard deviations of normal controls (sera obtained from four different healthy volunteers with blood group AB).
Immunohistochemistry and histomorphology
C4d staining was performed on paraffin sections applying a polyclonal rabbit anti-C4d antibody (Biomedica) (23). Biopsies were considered C4d positive in case of linear endothelial C4d deposition in at least a quarter of cortical peritubular capillaries. Allograft rejection was diagnosed according to the definitions given by the Banff 97 working classification of renal allograft pathology (24,25).
Continuous data are presented as the median and the interquartile range (IQR). The chi-square test was used to compare groups of categorical data and the Mann–Whitney U-test for comparison of continuous data. Kaplan–Meier analysis was applied to calculate graft survival. Statistical calculations were performed using SPSS for Windows, version 14.0 (SPSS Inc., Chicago, IL).
Posttransplant HLA reactivity in patients having a functioning graft for at least one 1 year
First, an overall cohort of 164 kidney transplant recipients, who survived with a functioning allograft for at least 1 year, was retrospectively evaluated for incidences and characteristics of posttransplant HLA alloreactivity. Patient demographics and clinical outcomes are described in detail in Tables 1 and 2.
Serial FlowPRA test results are shown in Figure 1A. Overall, at 2, 6 and/or 12 months posttransplantation, as many as 58 (35%) recipients tested positive by [IgG]FlowPRA HLA class I and/or class II screening (preformed [IgG]FlowPRA panel reactivity was detected in 36% of these recipients). Within the first year, 32 patients (20%) were positive with [C4d]FlowPRA HLA class I and/or class II screening (preformed C4d-fixing panel reactivity was found in 63% of these patients). As shown in Figure 1, [IgG] and [C4d]FlowPRA reactivity was most commonly observed at 2 months. Screening-positive patients were reevaluated by FlowPRA SA testing to specify targeted HLA antigens and to assess the presence of DSA. [IgG]DSA were found in 37 of the [IgG]FlowPRA-positive patients (64%) and [C4d]DSA in 12 of the [C4d]FlowPRA-positive recipients (38%).
A total of 111 recipients had biopsies for allograft dysfunction. Indication biopsies were performed at a median of 28 days (IQR: 16–293 days) after transplantation. Seventeen patients (15%) showed capillary C4d deposition. Sixteen of these recipients presented with typical morphological features of AMR (AMR Banff types: I: N = 10, II: N = 5, III: N = 1; chronic transplant glomerulopathy: N = 4). In accordance with our previous results (6), [C4d]FlowPRA reactivity was found to be associated with the occurrence of C4d-positive graft dysfunction. Seven of the 17 C4d-positive (41%), but only 17 of the 94 (18%) C4d-negative recipients tested positive by [C4d]FlowPRA (p = 0.03). According to the Banff scheme, AMR was confirmed for eight patients who showed a positive posttransplant [C4d] and/or [IgG]FlowPRA as well as typical biopsy-based features (C4d in peritubular capillaries and morphologic AMR criteria). Seven of these recipients had been subjected to immunoadsorption therapy according to previously described protocols (21,26). Remarkably, among biopsied patients, 38 were [C4d] and/or [IgG]FlowPRA positive without capillary C4d deposition and, with the exception of eight patients (capillary margination or microthrombi: N = 5; chronic transplant glomerulopathy: N = 3), without typical morphology. In a clinical comparison of these 38 recipients with the 8 patients having proven AMR no significant differences with respect to median eGFR (52 versus 69 mL/min, p = 0.7), urinary protein excretion (0.23 versus 0.12 g/24h, p = 0.2) or graft survival (87% versus 75%, p = 0.5) were found.
Posttransplant HLA reactivity in patients with an excellent 1-year clinical course
To evaluate if HLA alloreactivity also occurs in patients with an uneventful posttransplant course, we analyzed a subgroup of 34 kidney transplant recipients who, within the first year, have maintained excellent graft function without any episode of graft dysfunction, which presumably could have been mediated by humoral alloreactivity. This patient group was defined according to stringent clinical criteria, i.e. (a) an eGFR ≥60 mL/min and 24-h urinary protein excretion ≤0.5 g at 12 months, (b) no episode of graft dysfunction (and no indication biopsy) during the first 12 months and (c) to preclude an effect of interference with antibody levels, no application of peritransplant immunoadsorption for recipient desensitization. Baseline characteristics of this patient subgroup are described in Table 1. A comparison with the group of 130 recipients, who did not fulfill the above criteria of excellent 1-year graft function, revealed a lower donor age (p = 0.001) whereas for other baseline characteristics no significant differences were detected. Graft outcomes are summarized in Table 2.
The results of serial posttransplant FlowPRA screening (2, 6 and 12 months) are shown in Figure 1B. Within the first 12 months, 9 (27%) of the 34 recipients with excellent 1-year course tested positive by [IgG]FlowPRA (preformed reactivity was detected in four patients) and five recipients (15%) by [C4d]FlowPRA screening (preformed C4d-fixing reactivity was found in four recipients). As shown in Figure 1B, proportions of screening-positive patients within the group of patients with excellent 1-year course did not significantly differ from those found for the other 130 patients (Figure 1B).
For detailed characterization of alloreactivities detected among patients with excellent 1-year course, prescreened samples were further specified by FlowPRA SA testing and FCXM (Table 3). Applying FlowPRA SA testing, seven of the nine screening-positive patients were found to have reactivity against at least one of the HLA-coated beads represented in the SA panel. For two recipients with marginal panel reactivity (11%[IgG]FlowPRA reactivity) no [IgG]SA reactivity was detectable. Five of the SA-positive recipients had [IgG]DSA, whereby two had DSA with C4d-fixing capability (Table 3). Proportions of DSA-positive patients were not significantly lower than those detected for the other 130 recipients ([IgG]DSA: 15% versus 25%, p = 0.2; [C4d]DSA: 6% versus 8%, p = 0.7). Moreover, DSA-positive patients were evaluated for DSA binding strength (for recipients with two or more targeted donor HLA class I and/or II antigens, donor-specific reactivities with highest fluorescence intensity were included in the calculation). This analysis revealed no difference in average DSA binding between the 5 [IgG]DSA-positive recipients with an excellent 1-year course and the 32 [IgG]DSA-positive patients who did not fulfill the respective criteria [median levels of fluorescence intensity: 9.9 (IQR: 6.4–16.0) versus 8.3 (6.5–13.3), p = 0.5]. Finally, the nine [IgG]FlowPRA-positive patients with an excellent 1-year course were retested by cell-based alloantibody detection. This analysis revealed a positive T- and/or B-cell FCXM in three of the five patients for whom solid-phase DSA have been identified. Remarkably, one recipient without detectable [IgG]DSA (and SA) turned out to test positive by both T- and B-cell FCXM (Table 3).
Table 3. Antibody monitoring and clinical outcomes in antibody-positive recipients with excellent 1-year graft performance
Posttransplant antibody detection
Kidney allograft function
Protein excretion (g/24 h)
DSA = donor-specific antibodies; eGFR = estimated glomerular filtration rate; FCXM = flow cytometry crossmatch; no DSA = no detection of donor-specific antibodies; no SA = no detection of single-antigen reactivity.
1Test results are given as the maximum %PRA level detected within the first year (FlowPRA SA test results are listed in brackets).
2Proteinuria due to de novo membranous glomerulonephritis (maximum level of urinary protein excretion: 8 g/24 h at 3 years).
3Death with functioning graft due to disseminated pancreatic cancer.
42% (no DSA)
12% (no DSA)
12% (no SA)
46% (no DSA)
34% (no DSA)
11% (no SA)
11% (no SA)
63% (no DSA)
27% (no SA)
Posttransplant HLA reactivity in stable patients—impact on long-term allograft outcomes
Identifying a considerable number of antibody-positive patients showing excellent graft function within the first year, we were interested in the impact of detected HLA panel reactivity on long-term allograft outcomes. Clinical outcomes reported for the nine antibody-positive stable recipients are detailed in Table 3.
Among the 34 patients with excellent 1-year graft function, long-term allograft function (eGFR) did not differ between [IgG]FlowPRA screening-positive and -negative patients (Figure 2). At 1 year, urinary protein excretion turned out to be even lower in [IgG]FlowPRA-positive patients (Figure 2). Notably, among [IgG]FlowPRA-positive patients, one developed large proteinuria after 3 years (maximum level: 8 g/24 h) associated with a decrease of eGFR from 113 (1 year) to 61 mL/min 6 years after transplantation. A renal biopsy revealed de novo membranous glomerulonephritis (MGN) without any morphologic and immunohistochemical features of acute or chronic cellular or antibody-mediated rejection. Among [IgG]FlowPRA-negative patients, two developed proteinuria exceeding 2 g/24 h at 4 and 5 years, respectively, while maintaining stable graft function. One patient was subjected to biopsy which revealed C4d-negative glomerulitis with mild features of transplant glomerulopathy. In the other recipient, proteinuria spontaneously resolved to a level below 1 g/24 h. For both IgG-positive and -negative recipients no allograft loss was reported, however, within each subgroup, one patient death due to malignancy (disseminated pancreatic cancer and disseminated cutaneous squamous cell carcinoma, respectively) occurred.
As illustrated in Figure 2, recipients with C4d-fixing panel reactivity had eGFR levels comparable to those having no such reactivity. Furthermore, all five [C4d]FlowPRA-positive patients maintained 24-h protein excretion equal or below 0.5 g/24 h. At 1, 3 and 4 years, [C4d]FlowPRA-positive patients had significantly lower levels of protein excretion than [C4d]FlowPRA-negative recipients (Figure 2).
In the present study, we found that HLA alloreactivity may also occur in kidney transplant recipients with an excellent clinical course. A major finding was that, in the context of normal graft function, detected humoral alloreactivity did not predict inferior allograft function in the long term (median follow-up of 69 months).
Nine recipients (27%) with excellent 1-year graft performance were found to test positive by FlowPRA screening. In seven of these patients, HLA antibodies could also be identified by FlowPRA SA testing. This was not the case for two recipients, where weak panel reactivity in the absence of detectable SA reactivity can be speculated to reflect false positive FlowPRA screening. Notably, one of these SA-negative recipients turned out to test positive by FCXM. The possibility of positive crossmatch test results in the absence of solid-phase HLA alloreactivity is well established, and, for such cases, a role of non-HLA antibodies was speculated (27).
Importantly, among patients with excellent graft function, proportions of FlowPRA screening-positive recipients were not considerably lower than those found for patients with graft dysfunction. Moreover, no significant difference was observed with respect to DSA formation. Applying SA testing, for five of the nine antibody-positive stable patients circulating posttransplant DSA were identified. For three of these recipients, the presence of antidonor reactivity was reinforced by a positive T- and/or B-cell FCXM. For the DSA-negative recipients, HLA reactivity can be speculated to reflect irrelevant third-party reactivity. However, in this respect, it is important to note that undetectable DSA levels do not necessarily exclude an involvement of donor-specific alloreactivities. Indeed, some studies have demonstrated elution of DSA from rejecting grafts and have provided evidence that efficient antibody absorption by the transplant could impede DSA detection in the circulation (28,29). Profound antibody clearance by allografts is reinforced by studies demonstrating a substantial increase of circulating HLA antibodies following transplant nephrectomy (30,31).
Our present data indicate that posttransplant HLA reactivity may not invariably implicate inferior allograft function. Indeed, with the exception of one recipient, who developed de novo MGN without any features of antibody-mediated rejection, all antibody-positive recipients maintained stable long-term allograft function. Regarding the MGN case, a role of humoral alloimmunity can be speculated. Indeed, in a MHC-identical rodent transplant model, antibodies against an allogeneic tubular epithelial donor antigen have been suggested as a cause of transplantation-induced immune complex kidney disease (32). However, there is presently no convincing evidence for a pathogenetic role of anti-HLA antibodies in posttransplant MGN. In a recent anecdotal report, a patient diagnosed for de novo MGN was described to have circulating anti-HLA class II antibodies as a possible cause of disease (33). However, as in our patient, a pathogenetic relationship between alloantibody formation and MGN remains speculative as detected antibody could reflect coincidence, not causation.
In view of the intentional bias introduced by selecting stable patients according to strict clinical criteria, it is important to emphasize that our results should not be generalized to overall transplant cohorts or patient groups selected for graft dysfunction and/or rejection. In support of a major pathogenetic role of alloantibodies, numerous studies evaluating unselected cohorts or recipients subjected to indication biopsies have demonstrated tight statistical associations of anti-HLA antibody formation with allograft dysfunction and loss (1–6). In addition, experimental studies have reinforced a causal relationship between donor-reactive alloantibodies, intragraft complement activation and graft injury (34,35). In line with these results, a separate analysis of recipients subjected to indication biopsy revealed an association between [C4d]FlowPRA reactivity and C4d-positive graft dysfunction. Nevertheless, in accordance with a previous study (6), we identified a considerable number of antibody-positive patients who did not show any biopsy-based AMR features. These recipients did not differ from patients with proven AMR with respect to graft outcomes. Interpreting outcome data, however, it is important to note that most patients with proven AMR had been subjected to immunoadsorption. Considering the well-established antihumoral efficiency of extracorporeal therapy, it can be assumed that specific treatment in these recipients has ameliorated the natural course of rejection. This may impede a valid interpretation of clinical data in rejecting patients.
Our data are in accordance with few other reports suggesting that, occasionally, HLA reactivity might also occur in stable kidney transplant recipients. For example, in a recent cross-sectional study evaluating 198 kidney transplant recipients by Luminex HLA antibody screening, detection of anti-HLA class I and/or II alloreactivity (22% of the patients) was found not to be associated with inferior renal function (19). Moreover, in two other studies, DSA were reported to recur following successful desensitization by immunoadsorption without any impact on allograft function (15,16). Finally, in a recent multicenter study, Terasaki and Ozawa (36) reported that, in patients with functioning kidney grafts for at least 6 months, a single snapshot of HLA antibody detection was not predictive of subsequent graft failure among patients having serum creatinine values between 0.5 and 1.9 mg/dL. Even though neither serial antibody monitoring nor detailed data concerning long-term kidney allograft function were reported, the above mentioned study may support our conclusion that, in the absence of graft dysfunction, circulating alloantibodies may not necessarily implicate adverse graft outcome.
An interesting observation was that the five [IgG]DSA-positive patients with excellent 1-year course showed levels of DSA binding comparable to those observed for [IgG]DSA-positive patients with graft dysfunction or rejection. In some contrast to earlier reports suggesting an important clinical impact of DSA binding strength (37,38), our data suggest, that evaluation of antibody SA binding density may not always reliably distinguish between deleterious and less harmful alloreactivity. Moreover, applying [C4d]FlowPRA for detection of the complement-fixing ability of detected alloantibodies, four recipients were found to have C4d-fixing alloreactivity (two had [C4d]DSA). All four recipients presented with excellent long-term allograft function. The latter finding was surprising in view of our earlier observation that, in a patient cohort selected for acute graft dysfunction, detected C4d-fixing panel reactivity was tightly correlated with capillary C4d deposition and other morphological features of antibody-mediated rejection (6). In accordance with these results, in our overall population including patients with graft dysfunction, we also found a statistical association between [C4d]FlowPRA and the occurrence of C4d-positive rejection. However, since serial protocol biopsies were not available, an association of ex vivo detection of C4d-fixing HLA reactivity with in vivo complement activation cannot be thoroughly assessed. In this context, however, it is important to note that, in a recent protocol biopsy study, Mengel et al. (17) observed capillary C4d staining in some stable renal allografts. They found that, in this context, intragraft complement activation had no significant impact on allograft outcomes within a study period of 12 to 143 months posttransplantation (median 43 months) (17).
Within the stable subgroup most patients had reactivity against HLA class I antigens. However, only few recipients were FlowPRA HLA class II positive. In view of the earlier suggested prominent role of HLA class II reactivity as a risk factor for late graft failure (39,40), a low number of recipients with anti-HLA class II reactivity could have been in favor of superior long-term outcomes.
In a clinical context, our present data reinforce that the sole detection of circulating alloantibodies has to be interpreted with caution and may not justify implementation of specific therapeutic measures. This interpretation is in line with the Banff consensus, according to which the diagnosis of AMR should also be based on the detection of typical morphological features and capillary C4d deposits (25). Moreover, our results are in accordance with a recently published national consensus conference, where graft dysfunction was proposed as an additional diagnostic criterion (41). This is also supported by the recent observation that some kidney allograft recipients may maintain stable graft function despite C4d deposition and typical morphological features in protocol biopsies (17,42).
Interpreting our study, however, two major limitations have to be kept in mind, i.e. the small sample size and the lack of protocol biopsies, which were not routinely performed in our transplant cohort. Our clinical observation of excellent graft function in antibody-positive recipients may not preclude the development of mild subclinical morphological changes of AMR. Regarding the younger donor age observed for the stable group, it can be speculated that mild tissue injury has been compensated by the younger grafts over a prolonged period of time. This important issue will have to be clarified in a protocol biopsy study. A systematic evaluation of serial serological monitoring together with the results of protocol biopsies will be necessary to address the question of whether, in a therapeutic context, detection of alloantibodies in the absence of graft dysfunction should represent an indication for renal biopsy.
In summary, our study suggests that, in a considerable number of patients, posttransplant anti-HLA alloantibodies, also those with complement-fixing ability, may occur in the absence of graft dysfunction. Moreover, we found that, in the long term, such patients may not necessarily develop graft dysfunction, which, even though unproven, is tempting to speculate a role of accommodation also in the setting of ABO-compatible transplantation. Even though limited by its small sample size and a lack of protocol biopsies, our analysis suggests cautious interpretation of posttransplant serological monitoring in patients without early graft dysfunction.
This study was supported by grants from the Else-Kröner-Fresenius Stiftung to G.A.B. (project number: P37/06//A37/06) and to H.R. (project title: Anti-endothelial alloreactivity, 2000–2004).
Conflict of Interest Statement
The authors do not have any conflict of interest to report.