Posttransplant Major Histocompatibility Complex Class I Chain-Related Gene A Antibodies and Long-Term Graft Outcomes in a Multicenter Cohort of 779 Kidney Transplant Recipients

Abstract

Background 

The impact of major histocompatibility class I chain-related A (MICA) antibodies on renal graft outcomes is unclear. The goal of this work was to assess the impact of posttransplant MICA antibodies, assayed at 1 year, with two commercially available kits, on long-term renal graft outcomes.

Methods 

We retrospectively tested sera from 779 kidney transplant recipients with two single-antigen flow bead assays 1 year after transplantation. Samples were considered positive for MICA if they were positive in both tests or positive for MICA specificities that were present in one kit only. The main outcome was 4-year death-censored graft survival.

Results 

The prevalence of MICA antibodies was 5.4% at 1 year. MICA+ patients were more frequently human leukocyte antigen (HLA) sensitized and regrafted. Four-year death-censored graft survival was not different between MICA+ and MICA− patients (97% vs. 94%, P=0.28). By Cox multivariate analysis, independent risk factors for graft loss were as follows: number of HLA DR mismatches, acute rejection within the first year posttransplantation, 1-year serum creatinine, and the presence of HLA antibodies at 1 year, but not the presence of MICA antibodies.

Conclusions 

These data do not support an independent pathogenic role for MICA in long-term renal graft injury and question the interest of posttransplant monitoring of MICA antibodies with single-antigen flow bead assays currently available.

Antibody-mediated rejection (AMR) is a distinct pathologic entity with a worse prognosis than T-cell–mediated rejection in renal transplantation (¹). To date, major histocompatibility class I chain-related antigen A (MICA) seems to be the most robust non-human leukocyte antigen (HLA) polymorphic antigenic system that is able to induce an antibody response, and it is possibly involved in acute and chronic graft loss (^2–11). MICA is expressed within the normal kidney in both endothelial cells and podocytes (¹²).

The impact of preformed MICA antibodies on graft outcomes is still unsettled. The poorer 1-year overall graft survival reported in kidney transplant recipients harboring pretransplant MICA could not be replicated neither in a cohort of 425 kidney transplant recipients with 10-year follow-up nor among cardiac allograft recipients (^{13, 14}). There is ongoing research to determine whether the presence of MICA antibodies in the posttransplant setting is associated with worse outcome after renal transplantation (^{2, 6, 7, 10}). The prevalence of posttransplant MICA antibodies ranges from 9% to 20% and was found to be higher in patients who later lost their grafts (^{2, 7, 10}). For instance, in the largest reported cohort, the 1- and 4-year death-censored graft survivals (DCGSs) of patients free of HLA antibodies but having MICA antibodies were 10% to 15% lower than patients without HLA and MICA antibodies (¹⁰). Although suggestive, the association between posttransplant MICA and graft loss might be due to a number of confounders rather than indicate causality. Indeed, MICA antibodies were more prevalent among regrafted and HLA-sensitized patients, who were at higher immunological risk (¹³). Furthermore, no study has assessed the role of previous rejection on graft loss as another possible bias, as MICA may appear as a consequence of rejection without playing any further causal role in later graft loss.

To address these methodological shortcomings, we assessed the impact of posttransplant MICA antibodies, assayed at 1 year, with two commercially available kits, on long-term renal graft outcomes in a multicenter cohort of 779 patients.

RESULTS

Patient Characteristics

Demographic characteristics of donors and recipients, immunosuppressive therapy, and data on first year events are reported in Table 1. Median follow-up was 37.1 months (range 12.1–142 months) after transplantation. MICA+ patients (n=42) were significantly more frequently HLA sensitized at day 0 (panel reactive antibody >5%) and at 1 year posttransplantation, and more often regrafted and of African ethnicity in comparison with MICA− recipients. The higher immunological risk of MICA+ patients resulted in more frequent thymoglobulin induction and more use of steroids at 1 year. Acute rejection (AR) episodes at 1 year occurred in 24% and 16% of MICA+ and MICA− patients, respectively (P=0.19). One-year serum creatinine and dipstick proteinuria were comparable between groups. The characteristics of MICA+ patients (Table 1) that were positive with the One Lambda kit (n=59; Canoga Park, CA) or Gen-Probe (n=47; Gen-Probe Transplant Diagnostics, Stamford, CT) kit were similar to the group of 42 MICA+ patients.

TABLE 1: Characteristics of patients

Prevalence of MICA Antibodies

A total of 42 (5.4%) patients were sensitized against MICA at 1 year posttransplantation (2.1% were de novo antibodies). Twenty-nine patients (3.7%) were positive by both tests, whereas 13 patients were positive for MICA specificities only detectable with the Gen-Probe kit. When the two assays were considered separately, the prevalence of MICA antibodies was 6.0% with the Gen-Probe kit and 7.6% with the One Lambda kit. Conflicting results were seen in 48 patients (6.2%) as 30 patients (3.9%) positive with the One Lambda test were negative with Gen-Probe and 18 patients (2.3%) positive with Gen-Probe were negative with One Lambda.

Posttransplant HLA Antibodies

A total of 257 (33%) patients had class I or class II HLA antibodies at 1 year posttransplantation. Of these, 75 (29%) had HLA donor-specific antibodies (DSA). Anti-class I DSA were detected in 37 patients, whereas 43 patients harbored anti-class II DSA. A total of 24 (3.1%) patients had both HLA and MICA antibodies at 1 year after renal transplantation.

Survival Rates of Patients and Grafts

Survival was plotted from the time of testing, at 1 year. By November 30, 2010, 13 patients of 779 were lost to follow-up, 50 had lost their graft, and 33 died with a functioning graft. The causes of graft loss (n=50) were as follows: uncontrolled AR (n=14), chronic rejection (n=12), infection (n=11), recurrence of primary nephropathy (n=9), cardiac failure (n=2), toxicity from calcineurin inhibitors (n=1), and BK polyomavirus nephropathy (n=1). Four-year and 8-year patient survivals were similar in MICA+ and MICA− patients (4 years: 93% [95% confidence interval (CI): 76%-98%] vs. 95% [92%–96%] and 8 years: 93% [76%–98%] vs. 89% [84%–93%], respectively; P=0.68). DCGS in MICA+ and MICA− patients were 97% [95% CI: 83%–99%] vs. 94% [92%–96%] at 4 years and 69% [33%–87%] vs. 83% [77%–87%] at 8 years, respectively (P=0.29) (Fig. 1A). Altogether, only four MICA+ patients lost their grafts. Three of them had also HLA antibodies. As 57% of MICA+ patients concomitantly harbored HLA antibodies at 1 year, a factor that impacts graft loss, we compared DCGS between MICA+ and MICA− patients free of HLA antibodies at 1 year. Four-year DCGS was not statistically different between patients with MICA antibodies alone (n=18) and those without MICA or HLA antibodies (n=496) (100% vs. 96% [95% CI: 93%–98%]; P=0.67).

FIGURE 1: Kaplan-Meier curves of actuarial death-censored graft survival among the whole cohort according to (A) major histocompatibility class I chain-related A (MICA) antibody status and (B) human leukocyte antigen (HLA) status at 1 year posttransplantation. The number of patients at risk is indicated at yearly intervals.

By Cox univariate and multivariate analysis, the presence of MICA was not significantly associated with graft loss (Table 2). On the opposite, the number of HLA DR mismatches (hazard ratio [HR]=1.49; P=0.02), AR within the first year posttransplantation (HR=2.56]; P=0.005), 1-year serum creatinine (HR=2.38; P=0.009), and the presence of HLA antibodies at 1 year (HR=2.18; P=0.009) were independent risk factors for graft loss. DCGS among patients with HLA antibodies at 1 year was significantly worse compared with patients without HLA antibodies (4 years: 91% vs. 96% and 8 years: 72% vs. 88%, respectively; P=0.0008) (Fig. 1B).

TABLE 2: Cox multivariate analysis of death-censored graft loss after 1 yr

We next analyzed the primary outcome (DCGS at 4 years) in subsets of patients positive in one MICA assay only. Patients only positive with the One Lambda kit (n=59) or Gen-Probe kit (n=47) had a 4-year DCGS similar to patients without MICA antibodies (One Lambda: 98% vs. 94%, P=0.84; Gen-Probe: 98% vs. 94%, respectively, P=0.41). We also compared DCGS at 4 years in the 16 patients with de novo MICA antibodies at 1 year, and the 26 patients with preexisting MICA antibodies at transplantation. DCGS was not statistically different between both groups (87.5% vs. 96.2%; P=0.55). Furthermore, the rates of AR and chronic AMR (CAMR) were also comparable between both groups (AR: 6.3% vs. 3.8%, P=1.0; CAMR: 18.8% vs. 15.4%; P=1.0).

AR Episodes and CAMR

AR occurred in 15 patients (1.92%) after 1 year posttransplantation. Two MICA+ patients (4.8%) experienced a first AR episode after testing compared with 13 MICA− patients (1.8%) (P=0.19). AMR or suspicious AMR developed in one MICA+ and four MICA− patients.

CAMR occurred in 23 patients (3%) after testing at 1 year posttransplantation. The incidence of CAMR was significantly higher in MICA+ patients compared with MICA− patients (7 of 42, 16.7% vs. 16 of 737, 2.2%, respectively; P<0.0001). However, six of these seven patients also harbored HLA antibodies, among which four were DSA. By univariate analysis, the following parameters were associated with CAMR: presence of MICA antibodies at 1 year, recipient younger than 50 years, HLA sensitization before and at 1 year posttransplantation, re-transplantation, a history of AR episode within the first year after renal transplantation, and 1 year serum creatinine more than or equal to 1.5 mg/dL (Table 3). The low number of events (n=23) and the close association of MICA with HLA sensitization precluded us from performing any meaningful multivariable analysis.

TABLE 3: Risks factors for chronic rejection

DISCUSSION

Our first finding is a 5.3% prevalence of MICA antibodies at 1 year posttransplantation, similar to the reports by Terasaki et al. (8.9%) (¹⁰) and Cox et al. (7.5%) (¹⁵). Like Cox, who considered positive MICA sensitization when MICA antibodies had been found in two of three Luminex bead-based assays, our samples had to be positive for MICA in both commercially available assays except for those that could be detected with the Gen-Probe assay only.

In our series, 4- and 8-year DCGSs were similar between MICA+ and MICA− patients, despite that MICA+ patients were at higher immunological risk. DCGS was also equivalent when patients were positive for MICA in just one assay. The more frequent thymoglobulin induction and greater use of steroids at 1 year posttransplantation might play a role in the equivalent graft survival of MICA+ versus MICA− patients, despite the overall higher immunological risk of MICA+ patients. We must also acknowledge that the number of MICA+ patients beyond 4 years of follow-up was small, and that DCGS was numerically lower among MICA+ patients at 8 years (Fig. 1). However, the figure shows only a crude, univariate analysis that does not take into account the higher immunological risk of MICA+ patients because of higher percentage of HLA-sensitized, retransplants, and black recipients. Indeed, by multivariate analysis, graft loss was associated with HLA DR mismatches, AR within the first year, serum creatinine at 1 year more than or equal to 1.5 mg/dL, and the presence of HLA antibodies at 1 year, but not with MICA antibodies. Along this line, a major methodological problem in the study of MICA, in addition to their low prevalence, is the large proportion of patients with concomitant HLA antibodies, a well-known risk factor for graft loss as again shown in our cohort. Excluding HLA positive patients left us with only 2% of patients with MICA antibodies alone, and their DCGS was equivalent to those without MICA and HLA antibodies (at 4 years: 100% vs. 96%, respectively). We also examined the incidence of AR and CAMR as surrogate markers of long-term graft loss. AR was rather rare 1 year posttransplantation and was numerically but not statistically increased among MICA+ patients. Although the incidence of CAMR was higher among MICA+ patients, the low number of events (seven patients) and the close association with HLA antibodies (six of these seven patients, with four DSA) precluded us from confirming an independent role of MICA in our cohort. Although the pathogenic effect of MICA cannot be formally ruled out, the same methodological limitations (low prevalence, confounding by HLA antibodies) are likely to prevail in any other study looking at a role of MICA in graft loss.

In the only other series that reported survival data after renal transplantation, DCGS 4 years after a cross-sectional analysis of 1319 patients was 98% among patients free of both HLA and MICA antibodies, vs. 86% among those MICA+ (P=0.0001) (¹⁰). In that earlier cohort, most patients were grafted before the year 2000 (compared with after 2000 in this study), more frequently received azathioprine and cyclosporine (compared with tacrolimus and MPA here), and the Luminex assays were different. All these factors might contribute to explain the difference in DCGS. Of note, this study did not find a detrimental impact of MICA among recipients of living donor kidneys, a fact that dampens the robustness of the association between MICA and renal graft survival. The issue of the DSA nature of MICA antibodies has recently been investigated after renal and heart transplantation in two studies (^{14, 15}). MICA DSA were rather infrequent (ranging from 2.8% to 5.7%) in posttransplantation. In the absence of HLA antibodies, MICA DSA were not associated with cellular or humoral AR after kidney or heart transplantation. Furthermore, heart transplant recipients with MICA DSA had a trend to better 1- and 5-year graft survival than patients without MICA antibodies (^{14, 15}). Although we acknowledge that we do not provide data on the DSA nature of MICA antibodies, we believe that this question is somewhat futile as overall MICA posttransplant had no independent impact on outcomes. Therefore, their further resolution in DSA positive or negative is unlikely to bring meaningful information. Finally, we tested only a single moment in time, with no regular follow-up of patients. However, our design reflects the clinical reality, where patients are most often subjected to one assay after transplantation that, if positive, may require intervention.

In summary, these data do not support a meaningful pathogenic role of MICA, as detected by the commercially available assays, in long-term renal graft injury. It further shows that the study of posttransplant MICA sensitization is, on methodological grounds, severely hampered by the low prevalence of MICA+ patients, and their frequent association with HLA antibodies. Therefore, cohorts of thousands of patients will be needed to assess a possible independent role of MICA in the posttransplant setting. Whether this effort is medically relevant is questionable in view of the low prevalence of these patients.

MATERIALS AND METHODS

Patients and Specimen Collection

The sera of 779 ABO-compatible kidney transplant recipients, who underwent transplantation between January 1, 1999, and October 31, 2009, were provided through a collaborative consortium group consisting of two Belgian (C.U.B Erasme hospital and Cliniques Universitaires Saint-Luc, Brussels) and two French transplant centers (CHU Lille and Foch Hospital, Paris). All patients signed informed consent for serum sampling. The authors notify to adhere to the declaration of Helsinki. Patients selected for analysis were 18 years or older with a functioning graft at 1 year posttransplantation and had pretransplant and 1 year posttransplant serum available. Multiorgan transplants were excluded in this study.

Primary and Secondary Outcomes

The primary outcome of the study was DCGS 3 years after MICA assay (=4 years after transplantation). Data were recorded up to November 30, 2010. Graft failure was defined when a return to dialysis or preemptive retransplantation was necessary. Patients who died with a functioning graft or who were lost to follow-up were censored for DCGS analysis. Secondary outcomes were as follows: the occurrence of biopsy-proven AR episodes; biopsy-proven CAMR; and patient and overall graft survival. A total of 307 renal graft biopsies in 268 patients were performed for cause. Biopsy specimens were evaluated by light microscopy and immunofluorescence for C4d. Histological lesions were scored according to the Banff 2007 classification (¹⁶). CAMR was considered in case of histological evidence, along with at least one of the following criteria: (1) graft dysfunction, (2) C4d-positive immunostaining, and (3) evidence of circulating HLA DSA. Intravenous methylprednisolone pulses followed by tapered methylprednisolone doses were used as first-line therapy for acute T-cell–mediated rejection. Thymoglobulin was used to treat corticosteroid-resistant AR and as first-line therapy for selected patients with severe AR (Banff IIb and III). Suspicious or proven acute AMR was treated with plasmapheresis/intravenous Ig±rituximab according to the centers’ policy.

Sample Size Calculation

When we designed the study, we considered that a DCGS lower by at least 5% 3 years after MICA testing would suggest MICA antibodies are clinically relevant in renal transplantation. The sample size needed to detect such a difference with a type I error of 5% and 80% power, considering a prevalence of MICA antibodies of approximately 10% in the posttransplant setting, was 1882 patients. After collecting the serum samples from four centers (n=779 patients), we performed preliminary analyses that revealed a prevalence of MICA of 5.4%. In addition, more than 50% of these MICA+ patients also harbored HLA antibodies. There is a possible confounding impact of HLA antibodies on survival; therefore, at least 7000 patients should be studied to reach ±170 patients that would be HLA−/MICA+. Although this would have allowed us to test our working hypothesis, we believed that it was unrealistic to set up this cohort; therefore, we decided to report our findings on the present cohort of 779 patients.

Laboratory Assays

MICA Testing

Serum circulating IgG MICA antibodies were identified in sera taken 1 year posttransplantation using Luminex beads bound to single recombinant MICA antigens provided in the LSA-MIC (Gen-Probe Transplant Diagnostics) and LABScreen MICA Single Antigen (One Lambda) products according to the manufacturer’s instructions. The Gen-Probe kit detects 28 MICA antigenic specificities, which cover approximately 94% of the antigens present in whites and approximately 98% of those present in African Americans. When using the One Lambda kits, we first used the LABScreen Mixed product as a screening test according to the manufacturer’s instructions. Samples that were positive with the LABScreen Mixed product and samples positive with the Gen-Probe kit but negative with the LABScreen Mixed product were consecutively tested with the LABScreen MICA Single Antigen assay according to the manufacturer’s instructions. LABScreen MICA Single Antigen kits detect approximately 90% of MICA antigens. A sample was considered MICA positive only if positive in both tests. In addition, we also considered as positive patients who had MICA specificities that were only detectable with the Gen-Probe kit, as they were absent from the One Lambda MICA antigen panel. Pretransplant sera of patients positive for MICA at 1 year were also assayed to determine whether MICA antibodies were de novo or preexisting.

HLA Testing

All 1 year posttransplant sera were screened for HLA antibodies using Luminex beads coated with HLA-A, -B, -Cw, -DR, -DQ, and DP antigens (Gen-Probe Transplant Diagnostics), which provided a positive or negative result. Identification of class I and class II HLA DSA was performed by Luminex analysis using sets of 97 beads (class I) and 73 beads (class II) (Gen-Probe Transplant Diagnostics), respectively. HLA typing of renal transplant recipients was performed bymolecular biology (Innolipa HLA typing kit; Innogenetics, Belgium). Pretransplant peak and current panel reactive antibody were determined by complement-dependent cytotoxicity.

Statistical Analyses

Demographic characteristics at 1 year posttransplantation were compared according to MICA status using Student’s t test for normally distributed continuous variables and the Mann-Whitney U test for continuous variables without normal distribution. For categorical data, comparisons were made using the Fisher’s exact test or Pearson’s chi-square test, as appropriate.

We searched for the possible impact of posttransplant MICA antibodies on long-term AR, CAMR, and graft loss. The association between MICA antibodies as the dependent variable and risk factors was tested for graft loss, by univariate and multivariate Cox proportional hazards model and for CAMR, by univariate logistic regression analysis. We have presented the odds ratios derived from the logistic regression, the HRs derived from the Cox model, and the P value corresponding to the Wald’s χ² test. Patient survival and DCGS was calculated by the Kaplan-Meier method and compared by the log-rank test.

ACKNOWLEDGMENT

The authors thank their clinical nurses, Ms. B. Borré, and N. Lietaer, for their invaluable help in collecting the DNA and serum samples.