Renal Failure and Transplant Unit, San Salvatore Hospital, L'Aquila, Italy, and Department of General Surgery, Section of Transplantation, Wake Forest University School of Medicine, Winston-Salem, NC, USA
We report here on a European cohort of 27 kidney transplant recipients displaying operational tolerance, compared to two cohorts of matched kidney transplant recipients under immunosuppression and patients who stopped immunosuppressive drugs and presented with rejection. We report that a lower proportion of operationally tolerant patients received induction therapy (52% without induction therapy vs. 78.3%[p = 0.0455] and 96.7%[p = 0.0001], respectively), a difference likely due to the higher proportion (18.5%) of HLA matched recipients in the tolerant cohort. These patients were also significantly older at the time of transplantation (p = 0.0211) and immunosuppression withdrawal (p = 0.0002) than recipients who rejected their graft after weaning. Finally, these patients were at lower risk of infectious disease. Among the 27 patients defined as operationally tolerant at the time of inclusion, 19 still display stable graft function (mean 9 ± 4 years after transplantation) whereas 30% presented slow deterioration of graft function. Six of these patients tested positive for pre-graft anti-HLA antibodies. Biopsy histology studies revealed an active immunologically driven mechanism for half of them, associated with DSA in the absence of C4d. This study suggests that operational tolerance can persist as a robust phenomenon, although eventual graft loss does occur in some patients, particularly in the setting of donor-specific alloantibody.
patients who stopped immunosuppression and experienced graft rejection
Organ transplantation has become the reference treatment for end-stage renal failure and there is compelling evidence for continuous improvement of both graft and recipient survival over the last decades (1). This success is largely due to a better control of the alloimmune response, improvements in surgical techniques and clinical management, and in the control of infectious complications and neoplasia. Nevertheless, transplant recipients still have a shorter life expectancy and poorer quality of life, which is largely directly or indirectly related to the side effects of immunosuppression (2–6). CNI-related nephrotoxicity is also a major long-term concern (7). Moreover, immunosuppression has been clearly associated with numerous risk factors, such as dyslipidemia, hypertension, diabetes, and obesity (8). There is thus a strong medical rationale for the development of less toxic (9) and more effective immunosuppressive agents as well as methods to identify patients who may benefit from a reduced immunosuppressive regimen before clinical transplantation tolerance can be achieved (10). Ever since the first attempts to induce tolerance, made by Billingham et al. in the 1950s (11), an impressive number of studies have been performed in rodents where tolerance to MHC-mismatched allografts can be achieved. According to the definition of tolerance in rodents, besides having a well-functioning graft without histological lesions of rejection, tolerant recipients accept a second graft of the same donor origin but reject a third-party graft. The demonstration of immune tolerance in humans fulfilling these same stringent criteria obviously cannot be made, and the relevance of in vitro cell unresponsiveness as a surrogate of in vivo reactivity is debated. However, an increasing number of patients who display good graft function in the absence of all immunosuppressive drugs for several decades have been described in the literature (12–21). These patients can respond to immune challenge and do not suffer from a general deficit in immune function or from opportunistic infections (15,22). In the context of liver transplantation, the percentage of selected patients who may display this phenomenon subsequent to drug withdrawal has been described at around 20%. Moreover, the phenomenon appears to increase exponentially with time, being exceedingly rare during the first 5 years of transplantation but becoming frequent after 10 years (unpublished data). Although the kidney is likely to be less susceptible to successful immunosuppressive drug withdrawal, there is now mounting evidence that kidney transplant recipients can also become operationally tolerant (14–21). We and others have previously defined this state based on the clinical criterion of stable graft function in the absence of any immunosuppressive treatment for at least 1 year (15).
The mechanisms involved in the induction and maintenance of this process of tolerance remain elusive. Although transcriptional profiling and phenotype studies have shown that the blood of operationally tolerant kidney patients displays several specific characteristics (16,17,20,23–25), no validated assays or predictors of operational tolerance are currently available for use in the clinic.
In this paper, by bringing together the largest cohort of operationally tolerant kidney transplant patients studied to date (27 cases) and comparing them with two control cohorts (kidney recipients under immunosuppression and patients who rejected their transplant after interruption of immunosuppression), we set out to determine the clinical characteristics linked to this state of tolerance and to assess the long-term evolution of these patients after immunosuppression weaning.
Patients, Materials and Methods
Patients and control groups
Clinical “operational tolerance” was defined as stable kidney transplant function with <150 μmol/L creatinemia and <1 g/day proteinuria in the absence of immunosuppressive drugs for at least 1 year (15). Twenty-seven patients fitting with this definition who had received a kidney transplant between January 1971 and April 2002 were enrolled in the study. The first objective was to compare operationally tolerant patients (Op-Tol) to a group of case-control patients under standard IS (IS-Controls). This first control cohort comprised 26 patients who had received a transplant between December 1972 and December 2003, under a standard immunosuppressive regimen. These patients were selected from the multicenter DIVAT database (Données Informatisées et Validées en Transplantation,http://www.divat.fr, N° CNIL 891735 version 2, August 2004) based on matching with Op-Tol patients for date of birth, date of transplantation and duration of their clinical stability similar to the time “Off immunosuppression” in the Op-Tol group. It was impossible to find a matched IS-control for one Op-Tol patient based on the selection criteria. The second objective was to compare the Op-Tol group to a control group of 27 patients who had received a kidney transplant between August 1986 and January 2008 and who had rejected their graft after immunosuppression withdrawal for medical decisions or non compliance (REJ-Controls).
Materials and methods
Anti-HLA immunization determination: The presence of pregraft anti-HLA class I and II antibodies was determined by micro-lymphocytotoxicity (LCT) on a selected panel of typed HLA donors or by ELISA. Postgraft anti-HLA identification was obtained either by multiplex screening (LAT-M; One Lambda, Canoga Park, CA, USA) or by LCT or ELISA, then confirmed using LAT-M. Anti-HLA-positive sera were tested for donor-specific antibodies (DSA) using the high definition Luminex Single antigen assay (Labscreen Single Antigen; One Lambda). Note that pregraft serum samples were only available for two patients (4 and 7) to perform this further method retrospectively.
IgG subclass determination: Single Antigen Bead assays were also used to determine complement binding or noncomplement binding DSA IgG subclasses. The PE-conjugated anti-human IgG was replaced in the protocol with mouse anti-human antibodies specific for IgG1, IgG2, IgG3 and IgG4 subclasses (SoutherBiotech, Birmingham, AL, USA). LABScreen negative control (NC) serum (One Lambda) as well as positive control (PC) and NC beads served as positive and negative test controls. Mean Fluorescent Intensity (MFI) above 1000 was considered positive.
Histological analysis: Graft biopsies were not performed on well functioning kidneys (IS-Controls and Op-Tol) for ethical reasons but were performed in five tolerant patients (cases 4, 5, 7, 9 and 13) who, over time, failed to meet the criteria of tolerance (significant graft dysfunction and/or increased daily proteinuria). Histology analyses were also performed on a transplantectomy sample from one patient who had refused a biopsy at the onset of graft dysfunction (case 10). Serial deparaffinized 4-μm sections stained for hematoxylin eosin safran (HES), periodic acid-Schiff (PAS), Masson's trichrome and Jone's colorations were scored by a pathologist according to the latest Banff classification (26).
Immunohistochemistry: The following anti-human antibodies were applied to deparaffinized 4-μm sections: Anti-CD4 (clone 4B12-1:50-Novocastra), anti-CD8 (clone C8/144B-1:50-DAKO), anti-CD20 (clone L26-1:250-DAKO), anti-CD68 (clone PGM1-1:200-DAKO) and anti-CD138 (clone MI15-1:100-DAKO). Stains were done on a DAKO Autostainer plus™ using an EnVision™ FLEX kit. Stained slides were evaluated in the most densely infiltrated areas by counting the number of positive cells in three high power fields (×400 magnification) and calculating the mean number of positive cells per sample. The results were semi quantitatively assessed as absolute percentages of positive cells relative to the whole inflammatory infiltrate (considered as 100%). C4d staining was performed with indirect immunofluorescence on frozen sections using a monoclonal anti-C4d antibody (Quidel) and assessed by Banff 2007 C4d scoring in peritubular capillaries (27).
Infectious complications: These parameters were obtained from the computerized database for IS-control patients and from a specific questionnaire addressed to all participating teams asking for the date and type of serious infectious diseases, CMV infection, herpes, zona and urinary tract infection (UTI).
Statistical analysis: Our objective was to determine whether operationally tolerant patients (Op-Tol) presented differences in demographic characteristics or clinical events before and after transplantation compared to 26 matched patients (IS-Controls) under immunosuppression and 27 patients who lost their graft following immunosuppression weaning (REJ-Controls). Qualitative parameters were compared between the three patient groups using a Xhi2-test adjusted with a Monte Carlo test. Quantitative parameters were compared using bilateral Mann-Whitney and Wilcoxon tests. Differences were considered statistically different for two-tailed p-values <0.05 (*).
The time between transplantation and return to dialysis (death censured) and the time between transplantation and the first negative event (return to dialysis or death with a functioning graft) were analyzed simultaneously for the group of operationally tolerant patients and paired patients under IS. Survival analysis was performed using the Kaplan–Meier method (28). However, no Kaplan–Meier statistical comparison was applied to compare the two groups because of the insufficient number of Op-Tol patients. Nevertheless, confidence intervals (95%) are indicated.
Demographic characteristics of operationally tolerant patients
The characteristics of the 27 Op-Tol patients are summarized in Table 1. Of note, because 10 patients have already been described (15), an update of their clinical status for five additional years is provided in the case reports (Appendix 1). At the time of the study, 19 of the 27 Op-Tol enrolled still met the clinical criteria of tolerance, while eight patients no longer did. Comparisons of per- and postparameters were made with the 26 matched IS-Controls and 27 REJ-Controls under long-term immunosuppression (Table 2A and B).
Table 1. Summary of immunological characteristics of the operationally tolerant patients
All Op-Tol patients n=27
Patients remaining Op-Tol n=19
Patients who lost Op-Tol n=8
48.1% (13) undetermined for 1
62.5% (5) undetermined for 1
Acute rejection episode
33.3% (9) undetermined for 1
31.5% (6) undetermined for 1
74% (20) undetermined for 2 Median 7±5 (range 2–33)
68.4% (13) undetermined for 2 Median 8±5,8 (range 2–33)
87.5% (7) Median 8±5,8 (range 2–33)
HLA A-B-DR incompatibilities
Median 3 ± 2 (range 0–5)
Median 2 ± 2 (range 0–4)
Median 3 ± 1 (range 0–2)
HLA DR incompatibilities
Median 1 ± 1 (range 0–2)
Median 0 ± 1 (range 0–5)
Median 2 ± 2 (range 0–1)
33.3% (9) undetermined for 4 anti-HLA class I alone for 2 anti-HLA class II alone for 2 anti-HLA class I and class II for 5
15.8% (3) undetermined for 4 anti-HLA class I alone for 1 anti-HLA class I and class II for 2
75% (6) anti-HLA class I alone for 1 anti-HLA class II alone for 2 anti-HLA class I and class II for 3
Postgraft HLA immunization before IS withdrawal
7.4% (2) undetermined for 2 anti-HLA class II alone for 1 anti-HLA class I and class II for 1
5.2% (1) anti-HLA class II alone for 1
12.5% (1) undetermined for 2 anti-HLA class I and class II for 1
Postgraft HLA immunization after IS withdrawal
29.6% (8) undetermined for 2 anti-HLA class II alone for 1 anti-HLA class II alone for 4 anti-HLA class I and class II for 3
31.6% (6) anti-HLA class II alone for 4 anti-HLA class I and class II for 2
25% (2) undetermined for 2 anti-HLA class II alone for 1 anti-HLA class I and class II for 1
Postgraft DSA immunization
14.8% (4) undetermined for 3 anti-DSA class II alone for 3 anti-DSA class I and class II for 1
15.8% (3) undetermined for 1 anti-DSA class II alone for 3
12.5% (1) undetermined for 2 anti-DSA class I and class II for 1
After IS withdrawal
IS withdrawal (years from transplantation)
Median 11±6 (range 1–20) undetermined for 1
Median 13±11 (range 1–20)
Median 9 ± 1.5 (range 7–12) undetermined for 1
Withdrawal through noncompliance
Tolerance duration (years)
Median 9 ± 3.8 (range 2–20) undetermined for 1
Median 9 ± 3.5 (range 2–20)
Median 9 ± 4.5 (range 2–15) undetermined for 1
Table 2. Statistical analysis of qualitative and quantitative clinical parameters of the operationally tolerant patients compared to a matched cohort of stable patients under standard immunosuppression
PRA class I
>4 HLA-A-B-DR incompatibilities
>1 HLA-DR incompatibilities
Acute rejection episodes
No Induction therapy
Donor age (years)
Cold ischemia (hours)
PRA class I/II
>4 HLA-A-B-DR incompatibilities
>1 HLA-DR incompatibilities
Acute rejection episodes
No Induction therapy
Donor age (years)
Recipient age (years) At transplantation time
Recipient age (years) At the time of arrest of IS
Cold ischemia (hours)
Among the 27 Op-Tol patients, 5 (18.5%) received a kidney from an HLA identical related donor: 3 from sisters (cases 12, 14 and 24) and 2 from brothers (case 8 and 21) with no twin transplantation. Operational tolerance in this cohort of 27 kidney transplant recipients was observed as a result of patient non compliance in the majority of cases (n = 19). In the eight other patients, the interruption of IS was a medical decision due to PTLD (n = 5), skin and renal cancer (n = 2) and CNI toxicity (n = 1). Immunosuppression (IS) was stopped at a median time of 11 ± 6 years posttransplantation (range 1–20). By comparison, the 27 REJ-Controls were also non compliant in the majority of cases (n = 21). In four patients, the interruption of IS was a medical decision. The reasons were undetermined for two patients (suspected noncompliance).
Op-Tol patients did not differ from IS-Controls in terms of recipient/donor gender and recipient/donor age (NS), with 69.2% male recipients and mean donor age 27 ± 15.1 years in Op-Tol patients versus 61.5% and 30.5 ± 12.3 years for IS-Controls (NS). No significant differences were identified in the number of HLA-A-B-DR incompatibilities > 4 or HLA-DR matching between the Op-Tol patients and IS-Controls. The median donor-recipient HLA-A-B-DR matching was 3 ± 2.25. The percentage of patients that had undergone more than 10 blood transfusions was higher in Op-Tol patients (20.8%) compared to IS-Controls (12%) but did not reach statistical significance. Most of the transfusions had been performed before 1990 with nonleukocyte depleted blood. REJ-Control patients were significantly younger than Op-Tol not only at the time of transplantation (24.6 ± 8.4 vs. 34.7 ± 16.3, respectively; p = 0.0211) but also at the time of immunosuppression withdrawal (28.4 ± 9.0 vs. 44.8 ± 16.7, respectively; p = 0.0002). These parameters could not be analyzed for IS-Controls since the patients were matched for date of birth and time of transplantation with the Op-Tol. No other significant difference was observed between the two groups.
Interestingly, the common difference identified between the Op-Tol group and both the STA- and REJ-Controls was the proportion of patients that had received an induction regimen. Whereas only 3.7% of REJ-controls and 21.7% of STA-controls had not received induction therapy, almost half of the Op-Tol patients (52%, p = 0.0001 and 0.04, respectively) had not received induction therapy (Table 2).
There were, however, no differences in the maintenance therapy prior to interruption of immunosuppression; almost all Op-Tol patients had received a maintenance regimen comprising steroids (CS) (25 of 27; 92%) and CNI (17 of 27; 62%). Ten patients (3%) had received a treatment with AZA/CS.
And Thirty-four percent of Op-Tol patients (9 out of 26, 1 undetermined) had experienced treated acute rejection episodes (BANFF not available at that time) before withdrawal of IS (50% of which were acute rejection episodes that were confirmed by biopsy for only four patients), a roughly similar proportion as in the IS-Controls (Table 2). REJ-Controls had a higher incidence of acute rejection prior their treatment withdrawal (69.2%, p = 0.0468).
Overall, Op-Tol patients thus appear to display the usual characteristics of their counterpart patients under maintenance IS, with the exception of less induction therapy at the time of transplantation. REJ-Controls were significantly younger than Op-Tol recipients, not only at the time of transplantation but also at the time of immunosuppression withdrawal.
Follow-Up of the Operational Tolerance State
A graphic illustration of graft function evolution for each Op-Tol patient is provided in Figure S1. In our cohort of 27 patients who met the criteria for Op-Tol at inclusion, 8 (i.e. 30%) subsequently presented graft dysfunction (simultaneous increase in creatinemia and proteinuria) during the survey period. Annex Table 1 shows the detailed characteristics for each patient up to the loss of operational tolerance. The median duration of “Operational tolerance” fulfilling the definition criteria in these eight patients was 10 ± 5 years (range 2–16), with only one patient (Case 7) losing stable graft function within the first 5 drug-free years. The 19 remaining Op-Tol patients continue to show stable function at the time of writing, with a mean creatininemia of 114 ± 47.6 μmol/L (range 62–152) and proteinuria of 0.20 ± 0.26 g/day (range 0 to 0.9). The median drug-free period for these patients is 9 ± 4 years. As such, our data suggest that the duration of this unique phenomenon is highly variable but can be as much as 20 years, with a median of 9 ± 4.8 years, which is quite remarkable.
To gain better insight into the actual duration of operational tolerance, changes in graft function throughout the drug-free period were carefully compared to the IS-Controls under immunosuppression. Patient and graft survival are illustrated in Figures 1 and 2, respectively. These data show that graft failure in the Op-Tol cohort did not occur earlier than in the matched IS-Controls.
We then looked in more detail at the status of the eight patients who subsequently lost their state of operational tolerance among the 27 operationally tolerant recipients. Considering the small number of patients in each subgroup (19 continued Op-Tol vs. 8 with loss of Op-Tol), demographic comparisons were not statistically appropriate. One can note, however, that the median HLA-A-B-DR and DR incompatibility appeared to be higher in the eight patients who presented with graft failure and that six out of the eight had been immunized before transplantation (anti-HLA antibodies; n = 1 anti-class I only, n = 2 anti-class II only and n = 3 both). In contrast, among the 19 patients whose graft function remained stable after IS withdrawal, only three patients had detectable anti-HLA pre-transplantation (Table 1, undetermined for four patients).
Immunization profiling in operational tolerance
Given the time that had elapsed since the transplantation in the three patient groups, several techniques had been used to determine donor and patient HLA typing and pre and postgraft immunization. Regrettably, it was technically impossible to apply the current method to the majority of the patients, which limits the interpretation of the results. As the technology to detect anti-HLA immunization has evolved, the techniques used have been mentioned in Table A1.
To determine whether ultimate graft failure in Op-Tol patients may have been related to humoral immunization, pre and postgraft anti-HLA immunization were determined, including donor-specificity, and compared with the two control groups. At the time of transplantation, 9/27 Op-Tol patients displayed pre-transplant immunization (two patients with anti-class I only, two with anti-class II only and five with both anti-class I and II). No difference was observed with the IS-Controls, suggesting no “general” nonresponsiveness to blood transfusion in Op-Tol patients. Interestingly, among the nine Op-Tol patients immunized before transplantation, six presented with graft failure following total immunosuppression weaning compared with only 3 of the 19 patients who remained stable (Table 1). Compared to the group of REJ-Controls, total class II pre-transplant immunization was significantly higher in the Op-Tol group (p = 0.0300), mainly in those Op-Tol patients that ultimately lost their “tolerance state” (Table 2B).
During the immunosuppression-free operational tolerance period, approximately 30% (8 out of 25 patients, 2 undetermined) developed anti-HLA antibodies. Four patients had anti-class II antibodies, of which three were DSA (Case 6 DQ7 21824, Case 16 DQ5 5739 and Case 18 DQ7 7704) (Table 4), 1 developed anti-class-I antibodies (non DSA) and 3 others developed both anti-class I and anti-class II antibodies, of which 1 was a DSA (Case 7, A31 10436 and DR53 12340). Among these eight patients with postweaning de novo immunization, six remained stable with good graft function (median 7 years postweaning). However, the other two patients lost their graft 4 years and 6 months, respectively, after the appearance of anti HLA immunization: Case 13 with anti-class I (non DSA) and Case 7 with both anti-class I and class II (A31 10436 and DR53 12340). There was no particular Ig sub-class specificity for the DSA as one patient had IgG1 (DQ7), one patient had IgG2 (DQ8) and two others had IgG4 (DR53, DQ5) (Table A4). Comparison with the REJ-control group showed that total class II pre- transplant immunization was significantly higher in the Op-Tol group (p = 0.03, Table 2A). Thus, pre-transplant immunization appeared to be linked more strongly to ultimate graft failure than posttransplant immunization after a period of operational tolerance.
Table 4. Diagnostic classification assay from operationally tolerant patients
Attributing a cause for graft loss subsequent to a period of operational tolerance
In order to better comprehend the underlying cause of graft function deterioration in previously operationally tolerant patients, histological analysis and immunophenotyping was performed (Table 3). Histology analyses were available for six of the eight patients with graft dysfunction. Three patients displayed transplant glomerulopathy (TG) and microvascular inflammation (g + ptc ≥ 2) with DSA (Table 3 Cases 4, 7, 10, Figure S2 g-a-e) and had a high number of graft infiltrating CD68+ cells (Table A3, Figure S2b) that may be attributed to chronic antibody mediated rejection despite the lack of C4d deposits. In addition to TG, the histology sample of Case 10 showed chronic allograft arteriopathy that could be due to antibody and/or cellular-mediated rejection (Figure S2g). For the remaining three patients, two (Cases 5 and 9) displayed mild interstitial fibrosis without evidence of rejection, no interstitial, tubular and microvascular inflammation for Case 9 and very mild and focal tubulitis with minimal interstitial inflammation (i1 – t1) without any tubular expression of HLA-DR and ICAM-1 (data not shown) for Case 5. None of the two patients had C4d deposition but one probably had DSA according to the presence of anti-class II anti DQ5, DQ6 and DQ7, but no donor HLA DQ typing was available for Case 9 (Table 3). The last patient (Case 13) presented a high grade of IF/TA without any evidence of a specific etiology (under the limitation of an adequate biopsy available for analysis (Figure S2f-h) but with interstitial inflammation and severe capillaritis (ti2 and ptc3) and non donor-specific anti-HLA antibodies (Table 3). C4d was not determined for this Case 13. Finally, 50% of patients (n = 3) had DSA and histological lesions of chronic active ABMR. The remaining three patients had nonspecific lesions that were either isolated (n = 1), or associated with anti-HLA immunization (likely DSA [n = 1] or non DSA [n = 1]).
Table 3. Histological diagnosis and Banff characterization of biopsies
Susceptibility to infectious complications in operational tolerance
It is currently unclear whether operationally tolerant patients are more susceptible to infections due to a poorly reactive immune system, or rather, less susceptible because of their lack of chronic exposure to IS drugs. In the population of 27 Op-Tol patients, there were no severe opportunistic infections. And 8 of 27 patients displayed bacterial or viral infections (pulmonary or urinary) after IS withdrawal: 3 pneumopathies, 1 arthritis, 1 abscess and 1 otorhinolaryngology problem (Table A2). In comparison, 15/25 (1 ND) of the IS-Controls displayed at least one infectious episode during the entire follow up, which were serious in certain cases: 6 pyelonephritis, 1 prostatitis, 8 pneumopathies, 1 encephalitis (Jc virus), 2 CMV disease. UTI (15/26), herpes and zoster infection (6 of 26) were more frequent. Op-Tol patients thus appeared to be less exposed to viral or bacterial infections than IS patients, sustaining the hypothesis that they are less susceptible because of their lack of chronic exposure to IS drugs rather than more susceptible to infections due to a poorly reactive immune system. Although no comparison was possible with a matched population of healthy volunteers, the data fit with the possibility of selective tolerance, as also suggested by the response of Op-Tol to flu peptide vaccination (29). In addition, we did not find the infections to be more frequent in Op-Tol patients who ultimately presented with graft dysfunction and there was no obvious clinical sequence closely associating infection with graft dysfunction over time.
Induction of tolerance in a transplant recipient remains one of the ultimate goals of transplant practitioners but has been met with many challenges in the clinic. In this paper, we focus on a particular type of tolerance in humans, operational tolerance, not occurring in a specific controlled setting but arising spontaneously in patients who, for one reason or another, stop taking their immunosuppression. This state of prolonged functioning of a transplant for decades without treatment offers a unique opportunity to analyze a paradigm of tolerance in humans. Several analyses (phenotyping, transcriptional profiling, functional studies) have been performed on some of these patients, as reported in Table A5, but to date, none have reported on their clinical and/or biological parameters. In this paper, we report on clinical correlates in 27 such recipients of a tolerated kidney transplant, 10 of whom we partially reported on previously (15) and for who we now provide long-term follow-up data. A detailed clinical review combined with benchmarking against two “control” populations of patients (a matched cohort of recipients with stable graft function under chronic immunosuppression and a control group of patients who stopped immunosuppression and consequently rejected their graft), aimed at better understanding this rather elusive phenomenon. Nevertheless, despite the information provided by the comparison between the Op-Tol and the two control groups, interpretation of the statistical results must take into account the small number of patients per group.
We first focused on the parameters relating to the transplant itself and particularly the parameters that were previously identified when analyzing the first cohort of 10 operationally tolerant patients (15). For this larger cohort of 27 operationally tolerant patients, there was no difference in donor age between the three groups. Interestingly, the recipients who had stopped immunosuppression and rejected their graft were significantly younger than the Op-Tol patients at the time of the transplantation and subsequently at the time of immunosuppression withdrawal, suggesting and supporting previous reports that younger adult recipients may be more immunologically responsive, a situation that may negatively influence tolerance induction. From these data, it seems that time posttransplant does not significantly differ between the patients who maintained tolerance (13 ± 11) and those in whom tolerance was transient (9 ± 1.5), a difference with tolerant liver recipients that is perhaps not that surprising since the mechanisms involved in kidney and liver transplant tolerance are thought to be different (30,31). We also identified some distinguishing features. And 18.5% of the Op-Tol cohort had received a fully HLA-matched kidney from a living-related donor, which is of course proportionally higher than in the population of IS patients in general. As an example, among the 6704 patients from the DIVAT network who have received a transplant since 1990 (Necker, Nantes and Nancy in France), only 296 patients (4%) received a kidney from an HLA-identical related sibling donor. This finding supports data from other cohorts of Op-Tol patients where a higher proportion of HLA-matched patients has been reported (17). Since the first case of operational tolerance described in 1954 between twin brothers (32), only very few cases of operational tolerance have been described in renal transplantation in the absence of genetic identity between the recipient and donor. This observation may be related to a higher level of confidence in patients receiving a living-related kidney, leading to an increased frequency of non compliance in the context of a low level of alloreactivity, although one cannot exclude a selection bias where a higher proportion of fully HLA-matched patients are long-term graft survivors. In addition, HLA-matched operational tolerance may develop in a context of ignorance rather that tolerance. The comparison of the Op-Tol cohort with the matched IS-Controls and REJ-Controls highlights the fact that a significantly larger proportion of Op-Tol patients did not receive any induction therapy (48% vs. 21.7% and 3.7%, respectively, Table 2). This observation may be linked to the higher proportion of HLA-matched patients or to a counter tolerogenic effect of the induction therapy in the group of Op-Tol patients compared to the two other groups. However, there was no difference in the number of HLA incompatibilities in the other non-HLA identical recipients of the three patient groups.
We next focused on the features of operational tolerance development and maintenance. The survival graphs showed that patient and graft survival as well as kidney function were similar in the tolerant cohort and the cohort receiving immunosuppression. Although it is difficult to make any conclusions on so few patients, one may hypothesize that operationally tolerant patients do not seem to reject their graft earlier than patients under immunosuppression. Note that 2 of the 27 Op-Tol patients died during the study: Case 9, who presented with graft dysfunction died from old age and Case 11 died from suspected secondary cardiovascular events while maintaining stable graft function. The main reason for immunosuppression withdrawal was non compliance and although approximately half stopped their treatment progressively, as described previously (15), analysis of this larger patient cohort revealed that this “modus operandis” is not an absolute pre-requisite for becoming operationally tolerant, since approximately one third of patients were known to have weaned abruptly. Given that concomitant or secondary events such as viral infection have been shown to have dual effects on host responsiveness, in some circumstances rendering immune cells resistant to tolerance induction, and in others allowing tolerance to develop (33,34), we additionally appraised such events in the Op-Tol population. We found that Op-Tol patients displayed a normal blood formula during their period of tolerance, with no signs of inflammation (undetectable CRP blood concentrations) and a usual number of significant opportunistic, frequent or severe viral or bacterial infections after withdrawal of immunosuppression compared to the matched cohort of patients under IS. This does not suggest nonspecific immunosuppression in patients with operational tolerance. This is further supported by the fact that approximately one third of patients displayed acute rejection episodes before IS withdrawal (not significantly different from IS-Controls). Moreover, we have previously reported that operationally tolerant patients respond “normally” (at least not differently from normal healthy individuals) to an immunological challenge such as flu vaccination (35). These patients displayed a transcriptional regulatory profile with over-expression of certain genes specific to Treg cells, suggesting active regulatory mechanisms rather than exhausted alloreactive cells (36). Finally, these patients had a higher number of B cells with a higher expression of costimulatory and migratory molecules additionally associated with a regulatory B cell profile (25). As such, these data do not point towards immune senescence and/or hyporeactivity as a mechanism of operational tolerance in this patient population as has been previously suggested in long-term liver transplant survival (37). This constitutes another possible difference in operational tolerance between kidney and liver transplants.
Another interesting and intriguing observation is the fact that about 30% (8 of 27) of Op-Tol patients presented with graft dysfunction over time and therefore no longer met the “accepted” definition of operational tolerance at the end of this long-term follow-up period. The durability of operational tolerance has been questioned in the literature, and the proportion and speed with which some patients may subsequently lose their graft function is not well appreciated. The reason for this loss of function remains to be elucidated. A first key point of analysis in this cohort was their immunization status, as immunization could be expected to prevent tolerance or to promote its breakdown. In this larger cohort of patients, we were unable to confirm the low level of panel reactive antibodies reported for the first 10 patients that had suggested these patients to be low responders to blood transfusions (15). As such, the total proportion of sensitized patients and their level of PRA before transplantation was no different from the control cohort of patients under classical IS, but appeared to be significantly higher than in the group of patients who rejected their graft after withdrawal of immunosuppression. Moreover, the majority of the Op-Tol patients who lost their state of operational tolerance displayed pretransplant immunization. In contrast, the deterioration of graft function in the eight patients did not appear to be related to the development of a de novo humoral response as only one of the eight patients presented DSA. However, serum samples from patients who presented with rejection and whose grafts were subsequently removed were not studied because the patients were lost to follow-up. It is possible that the allograft can mask the antibodies by absorbing them, as described by Adeyi et al. (38). Two of these patients whose graft function deteriorated 29 and 38 years after transplantation were not biopsied and their serum displayed no PRA. We thus have no evidence to prove a loss of tolerance for these patients.
Among the six patients who underwent a biopsy for renal dysfunction, three displayed microcirculation lesions with TG and DSA, but none were positive for C4d. Nevertheless, these lesions could be attributed to active immune-driven mechanisms according to the 2011 Banff reports stating that microcirculation infiltration is the most significant parameter of ABMR and that C4d and DSA are limited in their diagnostic capacity. The other three patients displayed signs of IF/TA. In one of them IF/TA was isolated without anti-HLA immunization but the two other patients had anti-HLA antibodies (non-DSA for one and suspected DSA for the other). At this stage it is impossible to say whether the state of tolerance had been ruptured in the patient with the isolated IF/TA who lost their graft 19 years after transplantation. This is, however, plausible for the two patients who presented IF/TA and anti-HLA immunization, despite no proof of active immunologically driven lesions on the biopsy.
Finally, it is also interesting to note that six of eight patients with long-term graft dysfunction displayed preexisting anti-HLA antibodies but with very low PRA levels for three of them (1-7% LCT [Case 9], 1- 4% LCT [Case 10], 1- 12 % LCT [Case 5]), and they lost their immunization after transplantation. One patient (Case 4) was 55% HD before transplantation but no sera was available after IS weaning, one patient (Case 7) was 5% HD and displayed DSA after weaning, and one patient (Case 13) was hyperimmunized (95% LCT) and likely remained immunized without DSA after IS weaning. This highlights the need for prescreening of patients who may be candidates for minimization of immunosuppression to enable selection of patients who are not immunized prior to transplantation.
This analysis is of course limited by several factors. First, the duration of stable graft function without immunosuppression of at least 1 year used for the definition of Op-Tol was arbitrarily chosen and lends itself to criticism. Nevertheless, this definition is clinically relevant to detect acute events despite 1 year probably being too short to detect subclinical lesions and long-term graft dysfunction. In addition, the number of operationally tolerant recipients is small and the fact that we did not identify other significant differences between the groups of patients does not preclude the existence of such differences between the three populations. Finally, it was of course impossible to intentionally match the group of patients who rejected their graft with the Op-Tol patients, which may cause some degree of bias in the study.
Operational tolerance is known to be a meta-stable phenomenon. However, to date little is known about the underlying cause of declining graft function and subsequent loss of tolerance. It clearly appears in this study that the loss of graft function in the eight patients was not related to infections and that in half of the cases it may have been related to suspected ABMR without evidence of C4d. This observation shows that although biopsy monitoring would have been useful, at least in some selected patients with pregraft immunization, it would not have elucidated the cause of tolerance breakdown for the 3 other patients.
Despite the fact that operational tolerance appears to be extremely long-lasting in our patients, these histological findings, together with the finding of pre- or de novo anti-HLA/DSA antibodies, clearly indicate that a state a true immune tolerance, as defined in rodents, is likely not achieved in all patients. Operational tolerance in humans thus remains a complex and multifaceted clinical concept. Furthermore, the immunological pattern associated with the loss of the operational status itself is not unique, with only 25% of patients displaying anti HLA immunization after the IS withdrawal and before the loss of operational tolerance. Thus, besides a state of “true” operational tolerance that may concern (if biopsy-proven) 13 of our initial cohort of patients (44%), who maintain a functional graft for decades, with no opportunistic infections (15), a physiological response (29) and no anti-HLA antibodies, these findings give credit to the idea that “transplant operational tolerance does not mean complete unresponsiveness of the immune system toward the graft but rather a lack of a destructive immune response toward it” (39) and highly suggest that “a state of quiescence of the transplanted organ, functioning without a destructive immune response” (40) may exist alongside “true tolerance”. On this basis, these patients may meet the definition of “partial tolerance” rather than “tolerance” (Table 4). This concerns 6 of our cohort of 27 patients (24%) who maintained a functional graft for decades, with no opportunistic infections, a physiological response and who displayed anti-HLA antibodies (1 with DSA). The lack of management of the humoral immune response developing due to IS drug weaning does not necessarily precipitate graft loss, at least in a short-medium term (total duration 10 ± 5 years) in these patients.
Finally, eight patients presented with graft dysfunction (30%) within a highly variable time frame mean time of 10.5 ± 6 years. Thus, the histological analyses, taking into account the very small number of cases, suggest that 50% of failed grafts present immune activity but with features that are insufficient to establish a clear diagnosis. However, for the other half of patients, one possible hypothesis to explain nonspecific histological lesions could be the natural history and physiological outcome of over-filtrating ageing kidneys over several years, as for the general population or patients under IS. Indeed, all the biopsies (except cases 7 and 9 without glomeruli) showed glomerular lesions that could be due to nephronic reduction, such as moderate glomerulomegally associated with segmental or global glomerulosclerosis for cases 4 −5 and 10. Long-term (often one to several decades) loss of glomerular mass is a well known physiological event that is precipitated in low nephron mass conditions such as transplantation (41). The extent to which this physiological event contributes to late graft loss is not known. Longer graft survival in operational tolerance compared to IS controls could also be related to a lack of chronic CNI toxicity.
To conclude, the data presented here show that operational tolerance in human transplant recipients is a complex phenomenon with an extremely heterogeneous fate. Due to the relatively small number of patients, our data are primarily descriptive in nature and have to be interpreted taking into account these limits. Nevertheless, we propose a classification tree for these patients according to several successive criteria (Table 4): graft function, presence or HLA abs (± DSA) and histological lesions. We suggest that a “partial operational tolerance” process, in the absence of a destructive response towards the graft, may take place beyond “operational tolerance” and that the patients who suffer from graft dysfunction may be tentatively classified as “loss of tolerance” or “partial tolerance” or simply as patients with “aging kidneys”, like the majority of patients with well-functioning grafts under immunosuppression. Finally, we resume the different studies performed on these patients at different time points (Table A5). The constitution of registries of this still rare patient population will facilitate the further elucidation of this complex phenomenon, and hopefully enable a greater proportion of transplant recipients to achieve operational tolerance in a more controlled and safe manner.
This research was funded by a grant from the “Foundation Progreffe” (Nantes, France) and ROTRF grant. This work is also part of a French Transplantation Research Network (RTRS) supported by the “Fondation de Coopération Scientifique” CENTAURE.
The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.