Four-year Follow-up of a Prospective Trial of HLA and MICA Antibodies on Kidney Graft Survival


* Corresponding author: Paul Terasaki,


In 2002, 1329 patients with functioning transplants were prospectively tested for HLA antibodies in the 13th International Histocompatibility Workshop. Four years after testing, deceased donor graft survival among 806 patients not having antibodies in 2002 was 81% compared to 58% for 158 patients with HLA antibodies (p < 0.0001) and 72% for 69 patients with MICA antibodies (p = 0.02). Hazard ratio (HR) using death-censored graft survival from multivariate analysis of HLA antibodies was 3.3 (p < 0.00001) and 2.04 for MICA (p = 0.01).

In the 14th Workshop, at 1 year follow-up, survival for 1319 patients receiving deceased donor grafts and no HLA antibodies was 96% compared to 94% for 344 patients with HLA antibodies (p = 0.0004) and 83% survival for 33 patients with MICA (p = 0.0005). HR from multivariate analysis: HLA antibodies was 3.6 (p < 0.00001) and 6.1 for MICA (p = 0.006). Twelve patients with donor specific antibodies tested by single antigen beads had a 1 year survival of 64% (p = 0.008), and 27 patients with non-donor specific ‘strong’ antibodies had a 66% survival (p = 0.0003) compared to 92% survival in those with no antibodies.

In conclusion, these two prospective trials, after 1 and 4 years, provided strong evidence that HLA and MICA antibodies are associated with graft failure.


Despite increasing sophistication of immunosuppression, half of all kidney transplants performed from deceased donors fail within 10 years (1), as revealed by data on transplants performed between 1999 and 2003 from 33,118 ‘standard criteria’ deceased donors. This statistic may not be an accurate indication of actual loss since it excluded grafts from 5943 ‘expanded criteria’ deceased donors that had a 33% 10-year graft survival rate. In fact, more than half of deceased donor kidney transplants are lost within 10 years. Survival rates for other organ transplants, such as heart (2) and liver transplants (3), are no better. Clearly, the relentless attrition of grafts after the first year of organ implantation remains to this day the major hurdle to the long-term success of transplantation. Aside from the impact on individual patients, and the agony of returning to the long waiting list as a sensitized patient, loss of transplant function results in a major cost to society. For instance, patients whose kidney transplants fail must be returned to dialysis at an annual cost of $67 506. The 1 year cost of maintaining a functioning transplant, however has been estimated to be $13 749 (4). If we multiply the difference by 5% (loss per year) of 128 000 functioning transplants in 2003 (5), savings for Medicare can be calculated to be roughly $344 million dollars per year.

Although many factors have been associated with chronic transplant failure (6,7), including unalterable factors such as senescence of the graft and the race of the recipients, antibody-mediated failure, is a factor that can be potentially altered. Our earlier prospective study of 478 patients with HLA antibodies, with 2-year follow-up, showed that the presence of HLA antibodies was associated with a 15.1% failure rate compared to 6.8% among 1753 patients without HLA antibodies (8). Herein we report our findings after 4 years of follow-up on these same patients.

MICA is a locus linked to HLA (9) which determines a polymorphic series of antigens similar to HLA. Antibodies against MICA antigens have been found in transplant patients (10,11) as well as in patients prior to rejection of their grafts (12). They have also been found in eluates from kidney grafts (13). The frequency of MICA antibodies in rejected grafts has also been reported (12,14) Here we report the findings of a second prospective study that begins to clarify the role of MICA locus antibodies.


The sera of 964 deceased donor and 365 living donor kidney transplant patients were tested in 2002 for HLA antibodies by 21 centers as part of the 13th International Histocompatibility Workshop. Only patients lacking HLA antibodies before transplantation and who had functioning kidney grafts for at least 6 months were selected for analysis. Testing for Class I and II HLA antibodies was performed at each center using cytotoxicity, ELISA and flow cytometry. Four years later, the centers were asked to provide follow-up on graft survival of their patients. Graft survival was computed with death censored, for multivariate analysis, using Stata 9.1 (StataCorp, College Station, TX, USA). Univariate analysis are given for overall graft survival with no exclusions for deaths. In the figures, the values for graft survival with death censored are also given. Kaplan and Meier survival curves using Stata 9.1 was utilized and statistical probability computed with the log rank test. Centers that provided information are listed in the Acknowledgements section. Since the MICA antibodies could not be detected in 2002, in the current study, sera from patients who did not have HLA antibodies were sent and tested retrospectively for MICA antibodies in our laboratory.

Two years later, in 2004, 22 centers participated in the 14th International Histocompatibility Workshop. Criteria for inclusion in the study were the same as in the prior workshop. For this study, all reagents were provided from one source: One Lambda, Canoga Park, CA. The tests used were: ELISA, flow cytometry and luminex. Nine centers provided sera from 1166 patients for MICA testing in our laboratory. Eleven centers sent their positive sera to our laboratory for further testing with single antigen luminex beads for donor specific antibody studies. The single antigen beads were from One Lambda, and tested according to the manufacturer's instructions. Antibodies were arbitrarily defined as ‘strong’ when the normalized Luminex flourescence intensity was 6000 or greater and ‘moderate’ when the value was between 1800 and 5999.

MICA testing

MICA-expressing cell lines were produced from the m-HMY2.CIR cell line, which expressed no HLA Class I and II antigens. We electro-transformed these host cells with vector plasmid, MICA cDNA (*001, *002, *004, *007, *008, *012, *017 and *018). cDNA's of MICA were modified with the coding sequences (exon 2–4), signal peptides (exon 1), transmembrane regions (exon 5), and cytosolic tails (exon 6 and 7) with G418 resistance. MICA expression of each cell line was confirmed with MICA monoclonal antibody. After lysis and isolation of MICA, the soluble antigen was coated on luminex beads for analysis. Luminex technology was utilized to detect the MICA antibodies. The Luminex fluorescence value from the MICA assay for each sera sample was normalized by subtracting the fluorescence value of normal serum control in the same test set. The normalized fluorescence value of 3000 or greater was considered to be a positive reaction to the MICA antigen bead.

Only those patients who did not have preformed HLA antibodies were included in the analysis. Since 24% of the patients in the 13th workshop were tested by cytotoxicity post-transplantation, we estimate that this was approximately the fraction tested pre-transplantion with cytotoxicity. The remainder of the patients was tested by solid phase assays. The fraction tested by cytotoxicity pre transplantation can be estimated to be lower for the 14th workshop, though an estimate cannot be supplied since all post-transplant testing was performed by solid phase assays provided by the workshop. Data from the two workshops were analyzed separately.


HLA antibodies, post-transplantation

Figure 1 shows the frequency of antibodies found among patients with functioning transplants according to each center. The first section shows results of HLA antibody testing from 11 centers that tested only for HLA antibodies. The average percentage of HLA antibodies found in centers that used ELISA tests was 15.9%, by flow cytometry, 24.1% and by Luminex, 25.7%. Among the nine centers that submitted sera for MICA testing (shown in the middle section of Figure 1), the HLA antibody frequency was 23.3%, MICA 5.4% and HLA + MICA 3.5%. The frequency of having either HLA and/or MICA antibodies in kidney transplants was 32%. For liver transplant patients of one center, the total frequency was 43%. Results from the three centers that provided data on heart transplant patients were pooled since the number of these patients at each center was small.

Figure 1.

Frequency of HLA and MICA antibodies found in transplant patients post-transplantation. The frequencies of antibodies found by each of the 11 centers that tested only for HLA antibodies are shown on the left hand panel according to the method of testing: E = ELISA; F = flow cytometry; L = Luminex. In the center panel are results from nine centers that provided sera for MICA antibody testing in addition to submitting HLA antibody data. The far right hand panel provides the frequency of antibodies in liver and heart transplant patients. Results of heart transplants from four centers were combined since the number per center was small.

When the results were analyzed as a function of time elapsed since transplantation, the overall frequencies of antibodies found did not increase or decrease with time, but remained relatively constant from 1 to 10 years (Figure 2).

Figure 2.

Frequency of HLA/MICA antibodies found in patients from 1 to 10 years post-transplantation. Note that although most of the patients had been tested 1–3 years after transplantation, the frequency of antibodies found in patients transplanted many years previously remained relatively constant.

Graft survival

Graft survival with univariate analysis is given with no exclusions (including deaths) in the conventional way. In addition, shown in the box inserts of each figure are graft survivals with deaths censored. The overall graft survival rate of patients who were tested in 2002 as part of the 13th Workshop was examined. The overall half-life of deceased donor transplants accumulated from all of the centers was 10.8 years, and the half-life of living donor transplants was 11 years (1). The outcome of transplantation was then examined for those with and without antibodies. Survival was plotted from the time of testing, since we wanted to measure the effect of antibodies after they had been detected.

As shown in Figure 3, the overall 4-year deceased donor graft survival of 964 patients was 77%. This is a failure rate of 5.7% per year, which is close to the overall failure rate one would expect in the UNOS registry. The 4-year survival of 806 patients without HLA antibody was 81%. In comparison, 158 patients with HLA antibody had a survival of 58% (p < 0.0001). Among 69 patients with MICA antibodies, the 4-year graft survival was 72% (p = 0.02). Shown in the box inserts of Figure 3 is the death censored graft survival corresponding to the overall 4 year survivals given above.

Figure 3.

Deceased donor graft survival of patientssince the time of testing. The overall deceased donor graft survival was 77% 4 years after testing. A marked difference in graft survival for those who did not have HLA antibodies and those with antibodies is apparent. Patients retrospectively shown to have had MICA antibodies also had a graft survival lower than those without HLA and/or MICA antibodies. Given in the box insert in this and following figures are the graft survival rates with death censored, together with the p-values.

Among those with living donor grafts, 275 patients without HLA antibodies had a 4-year graft survival rate of 78% (Figure 4). In contrast, 90 patients with HLA antibodies had 62% survival (p = 0.0008), while 21 patients with MICA antibodies only had a 80% survival (p = NS).

Figure 4.

Living donor graft survival of patient since time of testing. Patients with living donor grafts also showed a marked difference in graft survival among those with and without HLA antibodies. No difference in graft survival could be shown for the 21 patients with MICA antibodies.

The deceased donor patients from the 14th workshop, at 1 year follow-up are shown in Figure 5. Among the 1319 patients without HLA antibodies, the 1 year survival was 96%. Those with HLA antibody had a survival of 93% (p = 0.0004). Regarding MICA antibodies, patients without MICA antibodies had a 1 year survival of 96.8% compared to 82.7% for 33 patients with MICA antibodies alone (p = 0.0005). If deaths were censored, the graft survivals were slightly higher and the p-values were slightly more significant as shown in the insert of Figure 5.

Figure 5.

Graft survival of patients receiving deceased donor grafts in the 14th International Histocompatibility Workshop. A statistically significantly lower graft survival was noted both for patients who had HLA antibodies and MICA antibodies.

At 1 year follow-up, living donor grafts of the 14th workshop included 595 patients with no HLA antibodies whose survival was 98% (Figure 6). The 131 patients with HLA antibodies had a 92% survival (p = 0.0005), and 19 patients with MICA antibodies only had 100% 1 year graft survival (p = 0.4). Lack of significance for MICA antibodies could be the result of small numbers of patients with antibodies available for analysis, and the possibility that some of the MICA antibodies had been non-donor specific antibodies which had been present even before transplantation. Such antibodies unfortunately could not be detected since pre-transplant sera were not available for analysis of MICA antibodies.

Figure 6.

Graft survival of patients receiving living donor grafts in the 14th International Histocompatibility Workshop. For patients with living door grafts a significant effect was noted for those with HLA antibodies, but no deleterious effect was noted for the 19 patients who had MICA antibodies.

Donor specific antibodies

Data from 11 centers that provided sera from their HLA antibody positive patients for testing with single antigen beads is given in Figure 7. The 502 patients who did not have antibodies had a 92% 1-year graft survival rate, and this rate was used as a reference. Out of 74 who had antibodies, 12 had donor specific antibodies and 62 had non-donor specific antibodies. The patients with donor specific antibodies had a 1 year survival rate of 64%.

Figure 7.

Graft survival in patients with donor specific and non-donor specific antibodies. For this analysis, 74 patients were available for analysis using single antigen beads. The 62 patients who had non-donor specific antibodies were divided into those with strong, and those with moderately strong antibodies. Graft survival rates were compared against the survival of patients without antibodies.

Since these patients had been selected as those without preformed antibodies, most of the non donor-specific antibodies can be considered to have originated after the transplantation. There are currently no published or accepted methods or criteria for quantitating the amount of antibodies bound to Luminex beads or any flow assay for that matter. Not all the flow beads have the same amount of antigen coated on their surface and not all of the antigens behave in the same manner when purified and coupled to beads. As shown in Figure 7, if patients are further divided arbitrarily into those with ‘strong’ and those with ‘moderately strong’ antibodies by the fluorescence intensity, a clear distinction was seen in the 1-year graft survival. Carefully controlled experiments need to be performed to develop and document this approach to confirm this finding. The 1-year graft survival of 27 patients with non-donor specific ‘strong’ antibodies was 66%. The overall p-value by log rank test was <0.0001. Within a 1-year period, only ‘strong’ antibodies had a noticeable effect. Thus, both donor specificity and antibody ‘strength’ were important factors in predicting 1 year graft survival.

Cox multivariate analysis

Cox regression analysis shown in Table 1 was run separately for the two workshops. Graft survival was computed with deaths censored. In the 13th workshop, HLA antibodies had the strongest effect with hazard ratio (HR) of 3.3 with a p < 0.00001. MICA antibodies had a HR of 2.04 with a p of 0.01. African American recipients had a HR of 2.3 with p = 0.009. Factors that had a beneficial effect were Caucasian recipients, CsA + MMF and three centers that had contributed the largest number of patients.

Table 1.  Cox regression analysis of 13th and 14th workshop graft survival with deaths censored
variableTotalAffirmative casesHazard ratio95% CIp-value
  1. All patients did not have preformed HLA antibodies pre-transplant.

Workshop 13
HLA antibody (±)13292483.262.4–4.5<0.00001
Sex (M/F)13116681.140.8–1.60.4
MICA pos vs. HLA + MICA neg (±)1081902.041.2–3.60.01
African American (y/n)502612.281.3–4.00.009
Cauc (y/n)5023290.58.36–.930.03
CsA + MMF used (y/n)13291310.5.25–.970.02
CsA + Aza used (y/n)13291140.75.4–1.40.37
Tac + MMF used (y/n)1329850.59.28–1.30.15
Regrafts (y/n)1236551.68.91–3.10.12
Top 3 contributing centers (y/n)13296360.62.45–.860.003
Living donor (y/n)13293651.28.92–1.770.15
Workshop 14
HLA antibody (±)23894753.62.3–5.8<0.00001
Sex (M/F)238715470.87.5–1.40.59
MICA pos versus HLA + MICA neg (±)1924526.12.1–17.30.006
CsA + MMF used (y/n)23895660.5.25–1.00.03
CsA + Aza used (y/n)23895061.6.97–2.60.07
Tac + MMF used (y/n)23894830.99.54–1.80.97
Regrafts (y/n)23821831.6.75–3.30.3
Top 5 contributing centers (y/n)23899530.46.26–.830.006
Living donor (y/n)23897261.2.74–2.00.4

In the 14th workshop, again, HLA antibodies were the strongest factor with HR of 3.6 and p < 0.00001. MICA antibodies also was a significant risk factor with HR of 6.1 p = 0.006. Negative risk factors were: CsA + MMF and patients from five centers that contributed the largest number data. Regraft patients did not have a significant risk presumably since all patients in the study were patients who did not have pre-transplant antibodies.


We describe here the remarkable finding that 4 years after a single testing for antibody, kidney grafts in patients with HLA antibodies continue to fail at a higher rate than grafts in patients without antibodies (Figures 3 and 4). This indicates that the presence of HLA antibodies is associated with lower graft survival. These results, which emerged from the 4-year prospective study of the 13th International Histocompatibility Workshop that was initiated in 2002, show that patients with antibody have a lower graft survival. Retrospective studies published earlier also suggest that in some instances it will be years before the effect of antibody results in graft failure (12,15).

We also report here on the new 14th International Histocompatibility Workshop study which had a 1 year follow-up. Even 1 year after a single testing, the effect of antibody is clearly seen (Figures 5 and 6). This confirms the positive findings of the independent prospective study of the 13th workshop (16) based on an entirely different set of patients, largely from a different group of laboratories. On the basis of these two studies, it is now clear that HLA antibodies found post transplantation is highly associated with graft loss. In addition to the significance of HLA antibodies, the present study shows that MICA antibodies, found in patients with functioning transplants prospectively in the 1 year study and retrospectively in the 4 year study, also have an effect on graft failure, though the effect is not as strong as that for HLA. Considering other factors, Cox multifactorial analysis with deaths censored, showed that MICA is a significant factor in graft failure both for the 4 year study (p = 0.01), and for the 1 year study (p = 0.006). Living donor transplants with MICA antibodies was small in number (20), and could have been eliminated from the paper, but is being included in the univariate figures for completeness sake. We also note that lack of association with graft failure could be attributed to non-donor specific MICA antibodies present before the transplant

Unfortunately, testing for MICA antibodies was not possible prior to transplantation since methods for identifying the presence of MICA were unavailable 4 years ago. If MICA antibodies were pre-existing, and were donor specific, these antibodies would be expected to cause early acute rejection and failure, and would have been excluded from this study of patients who had survived for more than 6 months. On the other hand, if the antibodies were non-donor specific, they would be irrelevant to the outcome, and may have reduced the association of antibodies to failures. In other words, our inability to exclude pre-existing MICA antibodies would only diminish the probability of finding a positive effect of MICA antibodies on subsequent graft failure.

Prior retrospective studies have identified MICA as important in graft failure. For instance, MICA antibodies were found in eluates of kidneys which had been rejected (13). They were also found in the sera of patients who eventually failed at a higher frequency than in those who had functioning grafts (12). The fact that MICA antibodies were found pre-transplantation in 25% of 85 waiting list patients (13), suggests that even before transplantation, MICA antibodies may need to be considered, especially since they have been suggested as a cause of hyperacute rejections in the absence of HLA antibodies (17).

With respect to the antibodies that were detected post transplantation, in the unsensitized patients of our previous study, among 4763 patients from 36 centers, the implantation of a kidney graft resulted in an incidence of HLA antibodies of 20.9% (16). The present study, based on an entirely new set of patients from 20 centers, showed that 23% produced HLA antibodies. When the incidence of MICA antibodies was added to HLA, the frequency of patients with antibodies rose to 32% of 1166 patients at nine centers. We can conclude from these two extensive studies that nearly one-third of kidney transplant patients have antibodies after transplantation. MICA antibodies were present in 9% of 1166 patients post transplantation tested in the current study. This is much lower than the incidence of 22.7% in 66 transplanted patients reported by Zou et al. (13). Much of this difference can likely be attributed to the difference in the source of MICA antigens used for testing. Our tests were based on antigens isolated from recombinant human cell lines, whereas those used by Zhou et al. were from insect cells. The refolded characteristics of the antigens could well explain the difference in the reactivity of antibodies.

The development of HLA antibodies is not unique to kidney transplant patients. A study of liver transplants patients revealed a 43% frequency of HLA and/or MICA antibodies in 122 patients. Similarly in our earlier study, patients who had undergone heart, liver, or lung transplants, showed a similar frequency of HLA antibodies (16). When examined in terms of ‘time after transplantation’, we see that the percentage of patients who have antibodies is roughly the same from the first year after transplantation to even the 10th year after transplantation. Any small variation could be attributed to the low numbers of patients in each category. Thus, it is likely that as patients with antibodies are lost, new patients produce antibodies, keeping the total fraction of patients with antibodies relatively constant.

In the second prospective study, we were able to study the antibodies in greater detail, particularly with respect to donor specificity. This was made possible by the development in the intervening time of single antigen beads (18) which permit the clear identification of antibodies in the serum (19). In addition, the use of the luminex antibody detection system has allowed us to quantitate the strength of the antibodies using fluorescence intensity. We found that among patients with antibody that was donor specific and strong, 36% of the grafts failed in 1 year (Figure 7). Even if the antibody was not donor specific, but was strong in activity, 34% of the grafts failed in 1 year. Antibodies of moderate strength, however, did not significantly impact 1-year graft survival. It is not immediately clear why non-donor specific antibodies would be present at such a high frequency (84% of the 74 sera tested for donor specificity), or why they had a clear deleterious effect on the graft. If the antibodies were preformed due to transfusions or pregnancies, or were anamnestic antibodies to an earlier immunization, they would not be expected to affect graft survival. We have earlier postulated that donor specific antibodies can be bound on the kidneys, and in some instances, may not be circulating in the periphery (20). In other retrospective studies, 70% of chronically rejecting patients were shown to have donor specific antibodies (21). It seems clear that after nephrectomy of a failed graft, donor specific antibodies appear in circulation (22,23).

Nevertheless, the problem of what the ‘extra’ non-donor specific antibodies might be is still not solved. One possibility that we are considering is that the antibodies may be directed at a second unidentified locus antigen. The existence of this second locus has been postulated (24), and supported by results of cytotoxicity tests (25). The extra antibodies found in the current study, however, were detected by solid phase analysis based on purified HLA antigens. Another alternative explanation might be that the second locus is akin to the minor antigens described by the Goulmy group (26), which depend on the reaction with peptides included within the pocket of the HLA molecule itself. These antibodies may not necessarily be donor specific in the conventional sense of donor HLA antigen donor specific. It remains to be seen if antibodies reactive to the peptides within the molecule as has been noted in other systems (27) might act as histocompatibility determinants.

In conclusion, our two prospective studies on the graft survival following detection of post transplant antibodies have now shown that HLA antibodies and MICA antibodies is are at increased risk for subsequent graft loss. We recommend universal testing of kidney transplant patients for antibodies post transplantation, and careful monitoring of serum creatinine if antibodies are detected.


Supported by funds from One Lambda Inc. P.I.T. has a conflict of interest in being Chairman and a major shareholder in One Lambda, a company that makes HLA antibody testing kits.

We thank the following 13th International Workshop collaborators for their 3-year follow-up data and their patient sera for MICA analysis: Alexander Abdelnoor: Beirut; Laurine Bow: Hartford; Frank Christiansen: Perth; Deborah Crowe: Nashville; Susan Fuggle: Oxford; Ines Humar: Zagreb; Luiz Jobim: Porto Alegre; William LeFor: Tampa; Robert McAlack: Philadelphia; Narinder Mehra & Arundhati Panigrahi: New Delhi; Jorge Neumann & Tatiana Michelon: Porto Alegre; Edgar Milford; Charles Carpenter & Kathryn Tinckam: Boston; Elissaveta Naumova: Sofia; Charles Orosz & Ronald Pelletier: Columbus; Antonina Piazza & Domenico Adorno: Rome; Masaharu Sada: Osaka; Constanze Schonemann: Berlin; Gilbert Semana: Rennes; Antonij Slavcev: Prague; and Brian Tait: Melbourne.

We also thank the following 14th International Workshop collaborators for their antibody and follow-up data: Josephina Alberú: Mexico City; Ines Alvarez & Roberto Toledo: Montevideo; Helena Alvez, Manuela Monteiro, Jose Teixeira: Porto; Dominique Charron & Chantal Gautreau: Paris; Raffaele Conca: Turin; Benjamin Gomez & Francisco Monteon: Guadalajara; Malek Kamoun: Philadelphia; Pamela Kimball: Richmond; Takaaki Kobayashi: Nagoya; Pawinee Kupatawintu: Bangkok; Manuel Muro: Murcia; Douglas Norman: Portland; Francesca Poli & Bernadette Colombo: Milan; Raynald Roy & Francine Cossette: Quebec; Guido Sireci: Palermo; Ella Van den Berg-Loonen: Maastricht; Adriana Zeevi: Pittsburgh; and the following centers for their antibody data: Leszek Paczek & Michal Ciszek: Warsaw; Clara Alonso: La Coruña; Patricia Campbell: Edmonton; Hans Grosse-Wilde and Falko Heinemann: Essen; Stephen Leech: Philadelphia; Kazunari Tanabe & Hideki Ishida: Tokyo.

We also thank Dr. John Benfield for his suggestions on the manuscript.