Major Histocompatibility Complex Class 1 Chain- Related Antigen A Antibodies: Sensitizing Events and

Major Histocompatibility Complex Class 1 Chain- Related Antigen A Antibodies: Sensitizing Events and

Impact on Renal Graft Outcomes

Anne Lemy, ^1,6 Marc Andrien, ² Karl M. Wissing, ^1,5 Khadija Ryhahi, ² Aure´lie Vandersarren, ³ Judith Racape´, ⁴ Christine Heylen, ³ Lidia Ghisdal, ¹ Emine Broeders, ¹ Pierre Vereerstraeten, ¹

Michel Toungouz, ² and Daniel Abramowicz ¹

Background. Major histocompatibility complex class 1 chain-related antigen A (MICA) antibodies (Abs) have been associated with renal graft loss in one large cohort. The triggering factors for MICA Abs and their autologous or allogeneic specificity have not been well defined. More data on the impact of MICA on renal grafts outcome are needed.

Methods. We tested sera from 494 controls and 597 patients with chronic kidney disease (CKD) for MICA using Luminex.

Forty CKD MICA ⫹ patients were genotyped for MICA alleles to determine their auto- or allospecificity. We compared MICA ⫹ with MICA ⫺ renal transplant recipients with regard to acute rejection episodes and long-term survival.

Results. Blood transfusions, previous transplantation, and more than two pregnancies were independent risk factors for the presence of MICA Abs, as were CKD stage V status and male gender. Among the 40 genotyped patients, allo-Abs alone were present in 32 patients, both auto- and allo-Abs in 4 patients, and auto-Abs alone in 4 patients. When we compared MICA ⫹ with MICA ⫺ patients, the incidence of acute rejection episodes during the first year (10.2% vs.

12.8%), as well as 1-year creatinine and proteinuria, were similar in both groups. At 10 years, actuarial patient (97.8%

vs. 87.6%) and overall graft survival (76% vs. 72%) were similar between MICA ⫹ and MICA ⫺ patients.

Conclusions. In summary, (1) sensitizing events for MICA Abs are the same as for human leukocyte antigen Abs; (2) MICA Abs did not adversely affect renal graft outcomes in our cohort.

Keywords: MICA, Allo- and autoantibodies, Sensitizing events.

(Transplantation 2010;90: 168–174)

A ntibody-mediated rejection (AMR) after renal trans- plantation has been recently recognized as a distinct pathologic entity and has an overall prognosis that is worse than T-cell mediated rejection. The main targets for AMR are human leukocyte antigen (HLA) antigens. In addition, vi- mentin, angiotensin-II receptor type I, and major histocom- patibility complex class 1 chain-related antigen A (MICA) have been recently described as potential antigenic targets for AMR after renal transplantation (1–9). MICA appears today as the most robust non-HLA polymorphic antigenic system

capable to induce an antibody (Ab) response involved in acute and chronic graft loss. Epithelial cells, endothelial cells, skin-derived fibroblasts, keratinocytes, and monocytes may express MICA, whereas resting lymphocytes do not (10).

At the present time, there remain important gaps in knowledge regarding the epidemiology and specificity of

Ms. Vandersarren is an employee of Gen-Probe that markets the MICA lu- minex assay. Therefore, Ms. Vandersarren might have a potential conflict of interest with the content of the article; Ms. Heylen is an employee of Gen-Probe that markets the MICA luminex assay. Therefore, Ms. Heylen might have a potential conflict of interest with the content of the article.

1

Renal Transplantation Clinic, Erasme Hospital, Universite´ Libre de Brux- elles, Brussels, Belgium. Currently, Department of Nephrology, Univer- sitair Ziekenhuis Brussel, Vrije Universiteit Brussel, Brussels, Belgium.

2

Hemobiology Department, Erasme Hospital, Universite´ Libre de Bruxelles, Brussels, Belgium.

3

Gen-Probe Transplant Diagnostics, Heuvelstraat, Nijlen, Belgium.

4

Public Health School, Erasme Hospital, Universite´ Libre de Bruxelles, Brus- sels, Belgium.

5

Department of Nephrology, Vrije Universiteit Brussel, Brussels, Belgium.

6

Address correspondence to: Anne Lemy, M.D., Erasme Hospital, 808, Route de Lennik, Brussels 1070, Belgium.

E-mail: [email protected]

A.L. participated in the writing of the paper, the performance of the research and data analysis. She received research grants from the Renal Transplan- tation Clinic and has no conflict of interest; M.A. participated in the writing of the paper and data analysis and has no conflict of interest;

K.M.W. participated in the writing of the paper and data analysis and has no conflict of interest; K.R. participated in the performance of the re- search and has no conflict of interest; A.V. provided technical help for the assays of MICA Abs and MICA genotyping; J.R. participated in data analysis and has no conflict of interest; C.H. provided technical help for the assays of MICA Abs and MICA genotyping. L.G. participated in the writing of the paper and has no conflict of interest; N.B. participated in the writing of the paper and has no conflict of interest; P.V. participated in data analysis and has no conflict of interest; M.T. participated in the writing of the paper and data analysis and has no conflict of interest; D.A.

participated in the writing of the paper and data analysis and has no conflict of interest.

Received 18 January 2010. Revision requested 15 February 2010.

Accepted 8 April 2010.

Copyright © 2010 by Lippincott Williams & Wilkins ISSN 0041-1337/10/9002-168

DOI: 10.1097/TP.0b013e3181e228f7

168 | www.transplantjournal.com Transplantation • Volume 90, Number 2, July 27, 2010

MICA Abs, as well as their impact on acute rejection episodes (ARE) and graft survival. First, the prevalence in healthy sub- jects has only been determined in a single and relatively small cohort where MICA Abs were detected in 2 of 35 patients (7).

Second, the immunizing events triggering the production of MICA Abs such as transfusions, pregnancies, previous trans- plantation, or gender remain ill-defined. Third, the auto- or allospecificity of the MICA Abs has been defined in only few patients (3, 8, 10, 11) where they were found to be predomi- nantly allo-Abs. Finally, the impact of MICA Abs on 1-year renal graft survival has been investigated in only one large study where they were found to be detrimental (9).

In this study, we compared the prevalence of MICA Abs in sera from 494 healthy volunteers and 597 patients with uremia. We next searched for association between demo- graphic and clinical characteristics and the presence of MICA Abs. We investigated the auto- or allospecificities of MICA Abs among 40 patients who were genotyped for MICA alleles.

Finally, we evaluated the possible impact of MICA at trans- plantation on rejection episodes, graft function, and long- term graft survival.

MATERIALS AND METHODS Specimen Collection

Sera from 1091 patients were obtained from three separate hospital serum banks. First, serum samples from 494 workers from our hospital (referred to as “controls”) were provided by the Laboratory of Immunology. Second, we obtained serum samples from 597 patients with stage V chronic kidney dis- ease (CKD) from two sources. Sera from 425 patients with CKD who under- went transplantation between January 1, 1999, and May 31, 2008, in our center were obtained just before renal transplantation. The remaining 172 CKD sera originate from prevalent patients routinely dialyzed in our unit.

DNA samples were collected on the day of renal transplantation in 1 ethylenediaminetetraacetic acid-treated sample tube (3 mL). Extraction of genomic DNA was performed using a phenol-chloroform extraction method and the use of phase-lock gel tubes (12). DNA samples were stored at 4°C and were available for 93 kidney recipients.

The study was approved by the Université Libre de Bruxelles (ULB)- Erasme Hospital ethics committee. All patients and controls signed informed consent for DNA and/or serum sampling.

Laboratory Assays

Identification of serum circulating IgG MICA Abs was performed using Luminex beads bound to single recombinant MICA antigens provided in the LSA-MIC product (Gen-Probe Transplant Diagnostics, Heuvelstraat, Nijlen, Belgium) according to the manufacturer’s instructions. LSA-MIC is made up of 23 different Luminex beads to which single, purified recombinant MIC glycoproteins are conjugated. These 23 MICAs represent 85% of the most frequent alleles present in the general population. The beads were incubated with undiluted test serum samples and washed to remove unbound Ab. Anti- human IgG Ab conjugated to phycoerythrin was added and after incubation, the test sample was diluted and analyzed on the Luminex instrument. To determine if a bead was positive or negative for bound Ab, we strictly applied the manufacturer’s suggested cutoffs. Cutoffs were 1000 for adjusted value 1 and 3 for adjusted values 2 and 3. The bead was considered positive if ad- justed value 1 was more than 1000 and adjusted value 2 or 3 was more than 3.

The adjusted value 1 was generated by the subtraction of the background median fluorescence intensity (MFI) value from the raw MFI of each indi- vidual bead. The background MFI is the background noise due to bead vari- ation, which could be found on the lot-specific recording sheet provided with the kit. Adjusted value 2 was obtained by dividing the adjusted value 1 by the MFI of the calculated control. The calculated control is the raw MFI value of the lowest ranked antigen bead. Finally, adjusted value 3 was generated by

dividing adjusted value 2 by the relative amount of antigen on each bead as found in the lot-specific recording sheet.

Detection of circulating class I and II HLA Abs at transplantation was performed using Luminex technology (Gen-Probe Transplant Diagnostics) in patients who had MICA Abs to determine whether immunization against MICA was associated with HLA sensitization.

DNA Genotyping

DNA samples have been collected on the day of renal transplantation.

MICA genotyping was performed in CKD stage V patients positive for MICA Abs for whom DNA was available (n ⫽ 40). Genomic DNA was extracted using a phenol-chloroform method and phase-lock gel tubes. DNA-based MICA typing was performed with a sequence-specific oligonucleotide (SSO) assay developed by Tepnel Lifecodes that is based on the Luminex xMAP technology. Briefly, each of 62 MICA SSO probes was bound to 62 distinct fluorescent populations of Luminex microspheres, which are then blended.

Exons 2, 3, 4, and 5 of the MICA locus were amplified using MICA-specific primers. For each amplified product, one of the primers was labeled with biotin. The biotinylated polymerase chain reaction products were then hy- bridized to the SSO-microspheres and the resulting hybridized product was detected using a streptavidin-phycoerythrin conjugate. The captured streptavidin-phycoerythrin is detected using a Luminex instrument. The confrontation of patient’s genotype with their pattern of MICA Ab specificity allowed us to characterize the MICA specificity as autologous or allogeneic.

Demographic Characteristics and Outcomes of Transplanted Patients

MICA ⫹ patients at transplantation were compared with MICA ⫺ patients for demographic characteristics and outcomes after transplantation. The majority of patients received a quadruple immunosuppressive regimen com- prising an anti-IL2 receptor monoclonal Ab, a calcineurin inhibitor, myco- phenolic acid (MPA), and steroids as recently reported (13). We studied the occurrence of delayed graft function (DGF), biopsy-proven ARE during the first year, plasma creatinine and proteinuria at 1 year, and 10-year patient and graft survival (13). The survival data were recorded up to May 31, 2009.

The median follow-up for survival analysis was 41 months. The histological score of ARE was reported according to Banff 2007 (14). Rejection episodes were treated as described (13).

Statistical Analyses

The epidemiological data available for all controls (n ⫽ 494) and patients with CKD (n ⫽ 597) were age and gender. In addition, for 425 patients with CKD who were later transplanted, the following data were also available:

dialysis technique (hemodialysis vs. peritoneal dialysis), dialysis duration, number of blood transfusions, pregnancies, and previous renal grafts.

Among the 425 patients with CKD, we analyzed the possible associations between the demographic and clinical parameters listed above and the pres- ence of MICA Abs. We next searched within the whole cohort of 1091 pa- tients for the effect of CKD, age, and gender on the prevalence of MICA Abs.

Categorical data were compared by the Fisher’s exact test and numerical data

by Student’s t test. All risk factors for which the P value was equal or lower

than 0.10 in univariate analyses were entered in a multivariate logistic regres-

sion analysis as independent variables, the absence or presence of MICA

Abs being the dichotomic dependent variable. Patient survival and death-

censored graft survival were calculated by the Kaplan-Meier method and

compared by the log rank test. The following factors that are known to influ-

ence graft prognosis were studied: gender and age of the recipient and the

donor, order of the graft, number of blood transfusions, duration of the

dialysis period and HLA sensitization, cold and warm ischemia times, num-

ber of HLA mismatches, DGF, and immunosuppressive protocol. Further-

more, presence or absence of MICA Abs was also studied. Multifactorial

analysis of graft survival was analyzed according to the Cox’s method. The factors

that reached statistical significance (P ⬍ 0.10) in univariate analyses were stepwise

entered into the model in an ascending order. Only those factors for which

P value was lower than 0.05 were considered to be significant.

RESULTS

Prevalence of MICA Abs and Associations With Demographic and Clinical Parameters

First, we analyzed the cohort of 425 CKD stage V pa- tients (Table 1). By univariate analysis, male gender, younger age, hemodialysis (vs. peritoneal dialysis), transfusion, preg- nancies, previous graft, and the presence of HLA sensitization were all associated with MICA Abs with odds ratios (ORs) ranging between 1.4 and 3.0. Three parameters remained sig- nificant by multivariable logistic regression analysis. MICA Abs were significantly more frequent among patients who had received blood transfusions (OR ⫽ 1.9 [95% confidence interval: {CI} 1.0 –3.5], P ⫽ 0.048), those who had received a previous organ transplant (OR: 2.2 [95% CI: 1.2– 4.3], P ⫽ 0.016), and women who had two or more pregnancies (OR: 3.2 [95% CI: 1.1–9.1], P ⫽ 0.027). Next, we analyzed the prevalence of MICA Abs among the whole cohort of 1091 individuals (494 controls and 597 CKD stage V patients). The prevalence of MICA was higher among patients with CKD (14.9%) than in controls (4.7%; Table 2). MICA Abs were also more frequent among men than in women in the whole cohort (14.2% vs. 6.8%), and also in the separate groups of controls and patients with CKD (Table 2). Age was not asso- ciated with MICA Abs. By multivariate analysis, CKD stage V and male gender were associated with an OR of developing

MICA Abs of 3.1 (95% CI: 1.9 –5.0, P ⬍ 0.001) and 1.6 (95%

CI: 1.1–2.5, P ⫽ 0.023), respectively. However, the association between CKD and MICA could be biased toward a higher frequency among patients with CKD because there were more men among patients with CKD (61% of 597 patients, vs.

28% of 494 controls, P ⬍ 0.0001), possibly contributing to a higher prevalence of MICA among CKD stage V patients.

Other potential confounders were a history of transplanta- tion and of transfusion. Data for these variables were not available for our control population. For the present analysis, we decided to consider controls as free of previous transfu- sion and transplantation. The potential for misclassification was considered low in this population of healthy workers. In addition, misclassification for these risk factors in controls would weaken the association of interest between uremia and MICA in the analysis rather than strengthen it. With regard to pregnancies, the older age of women with CKD (mean, 52 year vs. 42 year in controls, P ⬍ 0.0001) that could be associ- ated with a higher number of pregnancies was probably offset by a lower fertility and pregnancy rate than among healthy controls. Therefore, we decided to leave this parameter out of the present analysis.

Logistic regression modeling showed that after adjust- ment for gender, transfusion, and transplantation, the OR of MICA in uremic versus control patients decreased from 3.10 TABLE 1. Risk factors for the development of MICA antibodies in patients with chronic kidney disease stage V

Univariate analysis

Multivariate analysis

N MICA ⴙ MICA ⴚ MICA ⴙ

(%) P OR (95% CI) P OR (95% CI)

Gender

Females 158 18 140 11.4

Males 267 41 226 15.4 0.25 1.41 (0.46–2.72)

Age (yr)

Mean ⫾ SEM 425 43.4 ⫾ 1.4 47.8 ⫾ 0.7 0.013

Dialysis technique

PD 56 4 52 7.1

HD 333 48 285 14.4 0.15 2.19 (0.75–8.70)

Duration (mo) (Mean ⫾ SEM) 389 40 ⫾ 5 41 ⫾ 3 0.89

Previous transfusions

No 214 20 194 9.3

Yes 198 39 159 19.7 0.003 2.38 (1.29–4.48) 0.048 1.9 (1.0–3.5)

Mean N ⫾ SEM 412 2.8 ⫾ 0.6 2.6 ⫾ 0.4 0.86

Previous pregnancies

0–2 104 8 96 7.7

⬎ 2 54 10 44 18.5 0.04 2.73 (0.89–8.49) 0.027 3.2 (1.1–9.1)

Mean N ⫾ SEM 158 3.1 ⫾ 0.6 2.1 ⫾ 0.2 0.072

Previous graft

No 351 39 312 11.1

Yes 74 20 54 27 ⬍ 0.001 2.96 (1.51–5.65) 0.016 2.2 (1.2–4.3)

HLA sensitization ( ⬎ 5% PRA)

No 364 46 318 12.6

Yes 61 13 48 21.3 0.074 1.87 (0.86–3.84)

MICA, major histocompatibility complex class 1 chain-related antigen A; PRA, panel reactive antibody; OR, odds ratio; CI, confidence interval; HLA,

human leukocyte antigen; PD, peritoneal dialysis; HD, haemodialysis.

to 2.19 (95% CI: 1.25–3.83). This indicated that the high prevalence of MICA in patients with uremia was only partly explained by the presence of confounding factors, that is more men, transfused, and previously transplanted patients among the CKD cohort. The adjusted OR remained highly significant (P ⫽ 0.006). We also assessed the effect of uremia on MICA prevalence by univariate analysis after exclusion of patients with a history of transplantation or transfusion to avoid confounding by these variables. Although the exclusion of patients with these risk factors reduced the observed effect of uremia on MICA Abs (OR ⫽ 2.17 [95% CI: 1.2–3.9], it re- mained statistically highly significant (P ⫽ 0.008). The ob- served effect of uremia on MICA Abs in this analysis was also similar to the multivariate analysis, suggesting that the logis- tic regression model was not affected by collinearity. After stratification for gender, the effect of uremia on MICA was equivalent in men (OR ⫽ 2.15) and women (OR ⫽ 2.21), indicat- ing that gender did not act as an effect modifier of the association between uremia and MICA. Therefore, our analysis indicates that uremia is a risk factor for MICA production that is indepen- dent of the other risk factors identified in our cohort.

When we considered the 89 CKD stage V patients positive for MICA, 59 (66%) had been exposed to one or more of the risk factors detailed above (pregnancies, transplantation, or transfu- sions), whereas no such exposure could be found for 30 patients (34%), all men. In other words, among 597 CKD stage V pa- tients, 5% had MICA Abs without obvious sensitizing events.

Characterization of MICA Abs and Relationship With Anti-HLA Abs

Among the whole cohort of 112 MICA ⫹ patients, 16 showed reactivity against one single MICA antigen, whereas the other 96 reacted against multiple MICA antigens. The

mean number of MICA antigens recognized by the patient’s sera was 5.5 (95% CI: 4.6 – 6.4) for the whole cohort, 3.4 (95%

CI: 2.3– 4.4) for the controls, and 6.0 (95% CI: 5.0 –7.1) for patients with CKD.

In 40 patients with CKD, we confronted their MICA genotype with their MICA antigenic specificities. The MICA allele frequencies found in our patient sample were 47% for MICA*008, followed by 27% for MICA*002, and 22% for MICA*004. These findings are similar to recently published data (15). Allo-Abs alone were present in 32 patients and auto-Abs alone were discovered in 4 patients. Both auto- and allo-Abs were found in four patients. Altogether, 8 of 40 (20%) of our patient cohort had auto-Abs. Their primary nephropathy consisted of chronic glomerulonephritis (n ⫽ 6), autosomic polycystic kidney disease (n ⫽ 1), and vascular renal disease (n ⫽ 1). None of the patient had clinical or sero- logical evidence of systemic lupus. We next analyzed the prev- alence of auto- or allo-Ab status according to the history of sensitizing events among these 40 genotyped CKD stage V patients. Auto-Abs against MICA were numerically more fre- quent among sensitization-free patients (4 of 12; 33%) when compared with those with a previous sensitizing event (4 of 28, 14%).

We next investigated the relationships between HLA and MICA Abs among the 425 patients with CKD who were later transplanted. As shown in Table 1, 61 patients (14.3%) had HLA Abs as detected by a panel reactive antibody more than 5% and 59 patients (13.9%) exhibited MICA Abs. Thir- teen of 425 patients (3%) had both HLA and MICA Abs. The prevalence of HLA sensitization was higher among MICA ⫹ when compared with MICA ⫺ patients (13 of 59; 22%, vs. 48 of 366; 13%, P ⫽ 0.074; Table 1). To study the associations between MICA and class I or class II HLA antigens sensitiza- TABLE 2. Prevalence of MICA antibodies according to age, gender, and CKD

status

Univariate analysis

Multivariate analysis N MICA

ⴙ MICA ⴚ MICA ⴙ

(%) P OR

(95% CI

) P OR (95% CI)

Whole cohort

Controls 494 23 471 4.7

CKD V

597 89 508 14.9 ⬍ 0.001 3.59 (2.2–6.04) ⬍ 0.001 3.1 (1.9–5)

Females 585 40 545 6.8

Males 506 72 434 14.2 ⬍ 0.001 2.26 (1.48–3.49) 0.023 1.6 (1.1–2.5)

Age (yr) (mean ⫾ SEM) 1091 48.4 ⫾ 1.4 47.6 ⫾ 0.4 0.58 Controls

Females 355 13 342 3.7

Males 139 10 129 7.2 0.1 2.04 (0.78–5.17)

Age (yr)(mean ⫾ SEM) 494 43.6 ⫾ 1.8 42.4 ⫾ 0.3 0.45 CKD V

Females 230 27 203 11.7

Males 367 62 305 16.9 0.1 1.53 (0.92–2.59)

Age (yr)(mean ⫾ SEM) 597 49.6 ⫾ 1.6 52.5 ⫾ 0.7 0.1

a

Major histocompatibility complex class I chain-related antigen A.

b

Chronic kidney disease.

c

Odds ratio.

d

Confidence interval.

tion, we tested the available sera from all 111 MICA ⫹ patients from the entire cohort of 1091 individuals with single, HLA- coated beads by Luminex. Altogether, 32 of 111 sera were positive for HLA class I or II Abs. This corresponds to 30 of 88 (34%) patients with CKD, when compared with only 2 of 23 (9%) controls. Among the 30 patients with CKD with both MICA and HLA Abs, 14 had Abs against class I only, 6 against class II only, and 10 patients against both class I and class II HLA antigens. We also studied the Luminex HLA sensitiza- tion of the MICA negative kidney recipient candidates. Sera from 356 of 366 patients were available for analysis. Sixty- three patients (18%) had HLA Abs, a lower proportion when compared with 34% (30 of 88) for MICA ⫹ CKD patients (P ⫽ 0.0012). Among these 63 patients, 21 had HLA-class I Abs only, 11 exhibited HLA-class II only, and 31 had both HLA-class I and II Abs.

Impact of MICA Abs at Transplantation on Graft Outcomes

Among the 425 CKD V patients who have been tra- nsplanted, 59 (14%) were MICA ⫹ and 366 (86%) were MICA ⫺ at transplantation (Table 1). A comparison of the demographic characteristics showed that, when compared with MICA ⫺ patients, those MICA ⫹ were more frequently men, were younger, had more often been transfused and pre- viously transplanted, and were more frequently HLA sensi- tized (Table 1). The proportion of living donors was similar among MICA ⫹ and MICA ⫺ patients (8.5% and 10.1%, re- spectively). Dialysis duration was similar in both groups (Ta- ble 1). The other relevant parameters are displayed in Table 3.

Cold and warm ischemia times, the number of HLA mis- matches, and the nature of immunosuppressive agents ad- ministered were similar in MICA ⫹ and MICA ⫺ patients.

With regard to posttransplant outcomes, DGF occurred in the same proportion of patients ( ⫾ 11%). The same was true for biopsy-proven ARE. Six of 59 MICA ⫹ patients (10.2%) vs. 47 of 366 MICA ⫺ patients (12.8%) developed ARE within the first year (P ⫽ 0.62). The histologic scoring of rejection was roughly similar in both groups (Table 3) although there was a trend for a more frequent occurrence of rejections with fea- tures of AMR among MICA ⫹ patients (5 of 6 vs. 19 of 47, P ⫽ 0.08). Most ARE occurred within 4 weeks in both groups (Fig. 1). At 1 year, mean serum creatinine and proteinuria were similar in both groups (Table 3). One-year overall graft survival was also equivalent among MICA ⫹ and MICA ⫺ pa- tients (94.9% and 94.1%, respectively). At 10 years, patient survival was 97.8% in MICA ⫹ vs. 87.6% in MICA ⫺ pa- tients (P ⫽ 0.09). Overall graft survival was, at 5 and 10 years, 84% and 76% for MICA ⫹ patients, and 79% and 72% for MICA ⫺ patients (Fig. 2). Multivariate Cox anal- ysis revealed that the only independent factor for graft loss in our whole cohort was previous transplantation: at 10 year, graft survival was 85.0% in recipients of first grafts (n ⫽ 351) vs. 60.4% in regrafted patients (n ⫽ 74; P ⫽ 0.005).

As retransplants were more frequent among MICA⫹ pa- tients (34% vs. 15% in MICA ⫺ patients, P ⬍ 0.001), we evaluated graft survival among first grafts only according to MICA status. Overall graft survival was, at 5 and 10 years, 88% and 88% for MICA ⫹ patients, and 80% and 74% for MICA ⫺ patients (Fig. 2).

DISCUSSION

The first finding from this study is a threefold higher prevalence of MICA Abs in patients with CKD when com- pared with controls (14.9% vs. 4.7%). The prevalence of MICA Abs in our CKD population was slightly higher than the 11.4% detected in the pretransplant sera of 1910 kidney recipients (9), possibly because we assayed 23 different MICA specificities, instead of 5 in the study by Zou et al. (9). Al- though part of the increased prevalence of MICA among pa- TABLE 3. Demographic characteristics and outcomes of transplanted patients

MICA ⴙ (N ⴝ 59)

MICA ⴚ (N ⴝ 366) P N° HLA A-B-DR mismatches

m (SEM)

2.6 ⫾ 0.2 2.6 ⫾ 0.1 0.67 Ischemia times

Cold (hr), m (SEM) 16.6 ⫾ 0.9 15.7 ⫾ 0.3 0.29 Warm (min), m (SEM) 32.4 ⫾ 0.9 31.7 ⫾ 0.5 0.60

Calcineurin Inhibitor use 0.65

Cyclosporine 10 69

Tacrolimus 46 282

None 3 15

Antiproliferative use 0.09

Azathioprine 0 4

Mycophenolic acid 56 320

Neither 3 42

Induction therapy 0.3

Anti-IL2 receptor mAb 33 228

ATG 15 69

OKT3 4 10

None 7 59

Delayed graft function 7 (11.9%) 41 (11.2%) 0.67 Number of first ARE 6 (10.2%) 47 (12.8%) 0.67 Histology of first ARE

Suspicious for AMR 1 5

AMR 1 4

Suspicious for TCMR (borderline changes)

0 10

TCMR

Banff 1 0 6

Banff 2 1 12

Banff 3 0 1

Features of both TCMR and AMR

3 9

Number of second ARE 0 5 ⬎ 0.99

Serum creatinine (1 yr) (mg/dL), m (SEM)

1.49 ⫾ 0.07 1.41 ⫾ 0.03 0.28 Proteinuria (1 yr) (g/d),

m (SEM)

0.33 ⫾ 0.25 0.22 ⫾ 0.08 0.66 Patient survival (10 yr) 97.8% 87.6% 0.09 MICA, major histocompatibility complex class 1 chain-related antigen A;

HLA, human leukocyte antigen; mAb, monoclonal antibody; ATG, antithy-

mocyte globulin; OKT3, anti-CD3 monoclonal antibody; ARE, acute rejec-

tion episodes; AMR, antibody-mediated rejection; TCMR, T cell-mediated

rejection.

tients with CKD is likely due to prior renal transplantation and transfusions, logistic regression analysis and analysis of subsets of patients with CKD free of transfusions and trans- plantation suggest that uremia “per se” is associated with a twofold increased prevalence of MICA immunization. Why CKD would trigger MICA Abs is speculative, but factors such as uremia-induced endothelial stress with increased MICA antigen expression, or the contact with dialysate fluids or filters could possibly play a role (16). MICA Abs were also more frequent among men than in women both in the whole cohort and in the separate groups of controls and patients with CKD, despite preg- nancy being an independent risk factor for the development of MICA Abs. Of note, similar trends were observed in two recent cohorts of renal transplant candidates (MICA ⫹ : women 9.7%, men 12%) and heart transplant candidates (MICA ⫹ : women 10.6%, men 13.2%) in (9, 17). Splitting the women according to their number of pregnancies in these two cohorts might have

revealed the same association between pregnancy and MICA oc- currence as the one we have found. At the present time, the origin of the influence of male gender on the presence of MICA Abs is unclear and deserves further investigation.

In this work, we identified previous transplantation, pregnancy, and transfusion as independent risk factors for the development of MICA Abs with ORs of approximately 2.

These risk factors are the same as those of HLA immuniza- tion. Indeed, we observed in our cohort that patients sensi- tized against HLA antigens were somewhat more likely to have Abs against MICA as well, consistent with a previous observation (9). As a matter of fact, 15% of our patients with CKD were sensitized against HLA or MICA, and 3% against both. We found that transfusion may trigger MICA sensitiza- tion, which was not the case in the report from Zou et al.. This discrepancy might be explained by the fact that we catego- rized patients as having received 0 or at least 1 red blood cell transfusion, whereas more than 5 transfusions were required for patients to be considered as “transfused” in the article by Zou et al. (9). Before the advent several years ago of extensive deleucocytation, red blood cells transfusions contained many white blood cells. Both monocytes and activated T cells ex- press MICA antigens, thereby leading ground to the trigger- ing of MICA allosensitization.

It is striking that nearly one third of MICA ⫹ CKD stage V patients had no identifiable immunizing event, suggesting other possible pathways for MICA sensitization. Further- more, confrontation between MICA Ab specificity and MICA genotype revealed that 20% of CKD stage V patients exhibited MICA IgG Abs that were autoreactive, a finding distinctly rare with respect to HLA Abs (18). Since the completion of our work, a similar proportion ( ⫾ 20%) of auto-MICA Abs has recently been found among patients awaiting a cardiac graft and in the posttransplant period after kidney graft (17, 19). Possible explanations for the occurrence of auto-Abs are the immunization against various microbial agents that have epitopes crossreactive with MICA antigens or the denatur- ation of the antigens during purification, exposing cryptic FIGURE 1. Cumulative incidence of first-year acute re-

jection episodes among MICA ⫹ and MICA ⫺ patients.

MICA, major histocompatibility complex class I chain- related antigen A.

FIGURE 2. Left: overall graft survival among the whole cohort of MICA ⫹ (n ⫽ 59) and MICA ⫺ (n ⫽ 366) patients. The

number of patients at risk is indicated at yearly intervals. Right: overall graft survival among recipients of a first kidney

transplant (MICA ⫹ , n ⫽ 39; MICA ⫺ , n ⫽ 312). The number of patients at risk is indicated at yearly intervals. MICA, major

histocompatibility complex class I chain-related antigen A.

epitopes that could lead to false-positive results in the Lumi- nex assay (18, 20). We must acknowledge that we did not confirm the auto-reactive nature by other techniques, such as binding to MICA-bearing cells or western blotting, but rather used the operational definition derived from Luminex. In this respect, it is important to note that we strictly followed the manufacturer’s instructions to determine the positivity of samples.

Using Luminex technology, Zou et al. reported a poorer 1-year overall graft survival in kidney transplant re- cipients harboring pretransplant MICA Abs when compared with those free of MICA (88% vs. 93%, P ⫽ 0.01). Unlike these authors, analysis of our cohort showed that patients with MICA Abs had a somewhat better overall graft survival than MICA ⫺ patients. This occurred despite the fact that MICA ⫹ patients were at higher risk for ARE and graft loss than MICA ⫺ patients because of younger age, more previous transplants, and more HLA sensitization. As multiple transplantation was the only independent factor affecting graft loss in our cohort, we also focused our analysis on first grafts. There again, MICA ⫹ patients had a better, although not significant, 10-year overall graft survival (88% vs. 74%, P ⫽ 0.25). When 10-year death- censored graft survival was considered, the presence of MICA had a negligible negative effect ( ⫺ 2%). Furthermore, the in- cidence of 1 year ARE, graft function, and proteinuria, all important surrogate markers of long-term graft function, were virtually identical between MICA ⫹ and MICA ⫺ pa- tients. Altogether, the present data strongly argue against a significant effect of pretransplant MICA on graft outcomes. A comparison of the immunosuppressive therapy given to Zou’s patients and ours suggest that the overall amount of immunosuppression was higher in our cohort. Indeed, in- duction therapy was used in one third of Zou’s patients, ver- sus the majority of the present cohort; the predominant calcineurin inhibitor used was cyclosporine A in the study by Zou et al., and tacrolimus here; and mycophenolic acid was administered to two thirds of Zou’s patients, versus virtually all of our patients. The impact of MICA Abs might be smaller in more heavily suppressed patients. Of note, a recent publi- cation revealed that cardiac allograft recipients with pre- formed MICA Abs had better 5-year allograft survival than their MICA-negative counterparts (17). Today, the evidence pointing to a direct pathogenicity of MICA Abs in renal trans- plantation is only circumstantial, mainly based on associa- tions between the presence of MICA Abs in the sera or kidney eluates with acute or chronic graft rejection (5– 8, 21). There is no animal model of MICA-induced rejection, nor have pu- rified MICA Abs been shown to trigger rejection in experi- mental animals. Obviously, more fundamental and clinical studies are necessary to better delineate the possibly complex influence of MICA immunization on acute and chronic rejec- tion episodes after solid organ transplantation.

ACKNOWLEDGMENTS

The authors thank our clinical nurses, Ms. B. Borre´, F.

Bernard, and S. Arias-Lopez for their invaluable help in collect-

ing the DNA and serum samples; and Drs. A. Le Moine and M.

Abramowicz for their critical reading of the article.

REFERENCES

1. Dragun D. Agonistic antibody-triggered stimulation of angiotensin II type 1 receptor and renal allograft vascular pathology. Nephrol Dial Transplant 2007; 22: 1819.

2. Carter V, Shenton BK, Jaques B, et al. Vimentin antibodies: A non-HLA antibody as a potential risk factor in renal transplantation. Transplant Proc 2005; 37: 654.

3. Amico P, Honger G, Bielmann D, et al. Incidence and prediction of early antibody-mediated rejection due to non-human leukocyte anti- gen-antibodies. Transplantation 2008; 85: 1557.

4. Sumitran-Holgersson S, Wilczek HE, Holgersson J, et al. Identification of the nonclassical HLA molecules, mica, as targets for humoral immu- nity associated with irreversible rejection of kidney allografts. Trans- plantation 2002; 74: 268.

5. Mizutani K, Terasaki P, Bignon JD, et al. Association of kidney trans- plant failure and antibodies against MICA. Hum Immunol 2006; 67:

683.

6. Mizutani K, Terasaki P, Rosen A, et al. Serial ten-year follow-up of HLA and MICA antibody production prior to kidney graft failure. Am J Transplant 2005; 5: 2265.

7. Mizutani K, Terasaki PI, Shih RN, et al. Frequency of MIC antibody in rejected renal transplant patients without HLA antibody. Hum Immu- nol 2006; 67: 223.

8. Zou Y, Heinemann FM, Grosse-Wilde H, et al. Detection of anti-MICA antibodies in patients awaiting kidney transplantation, during the post- transplant course, and in eluates from rejected kidney allografts by Luminex flow cytometry. Hum Immunol 2006; 67: 230.

9. Zou Y, Stastny P, Susal C, et al. Antibodies against MICA antigens and kidney-transplant rejection. N Engl J Med 2007; 357: 1293.

10. Zwirner NW, Marcos CY, Mirbaha F, et al. Identification of MICA as a new polymorphic alloantigen recognized by antibodies in sera of organ transplant recipients. Hum Immunol 2000; 61: 917.

11. Alvarez-Marquez A, Aguilera I, Gentil MA, et al. Donor-specific anti- bodies against HLA, MICA, and GSTT1 in patients with allograft rejec- tion and C4d deposition in renal biopsies. Transplantation 2009; 87: 94.

12. Sambrook J, Fritsch EF, Maniatis T. Molecular cloning: A laboratory manual [ed. 2]. Cold Spring Harbor, NY, Cold Spring Harbor Labora- tory, 1989.

13. Wissing KM, Fomegne´ G, Broeders N, et al. HLA mismatches remain risk factors for acute kidney allograft rejection in patients receiving quadruple immunosuppression with anti-interleukin-2 receptor anti- bodies. Transplantation 2008; 85: 411.

14. Solez K, Colvin RB, Racusen LC, et al. Banff 07 classification on renal allograft pathology: Update and future directions. Am J Transplant 2008; 8: 753.

15. Gao X, Single RM, Karacki P, et al. Diversity of MICA and linkage disequilibrium with HLA-B in two North American populations. Hum Immunol 2006; 67: 152.

16. Galli F. Protein damage and inflammation in uraemia and dialysis pa- tients. Nephrol Dial Transplant 2007; 22: v20.

17. Smith JD, Brunner VM, Jigjidsuren S, et al. Lack of effect of MICA antibodies on graft survival following heart transplantation. Am J Transplant 2009; 9: 1912.

18. Morales-Buenrostro LE, Terasaki PI, Marino-Vazquez LA, et al. Natu- ral human leukocyte antigen antibodies found in nonalloimmunized healthy males. Transplantation 2008; 86: 1111.

19. Suarez-Alvarez B, Alonso-Arias R, Bravo-Mendoza C, et al. Identifica- tion of epitopes and immunodominant regions on the MICA protein defined by alloantibodies from kidney transplant patients. Transplan- tation 2009; 88: S68.

20. Claas FH, Doxiadis II. HLA antibody detection and kidney allocation within Eurotransplant. Hum Immunol 2009; 70: 636.

21. Terasaki PI, Osawa M, Castro R. Four-year follow-up of a prospective

trial of HLA and MICA antibodies on kidney graft survival. Am J

Ttransplant 2007; 7: 408.