Since the first successful combined liver kidney transplant (CLK) in 1984, renal failure no longer constitutes a barrier to transplantation in patients with end-stage liver disease and irreversible renal failure (1, 2). The negative impact of preexisting donor-specific antibody (DSA) on outcomes in kidney transplantation is well established (3). On the other hand, the relevance of allosensitization in CLK remains controversial. Both clinical experience and experimental models have suggested that a liver allograft seems capable of removing circulating anti-donor antibody and preventing humoral rejection of other allografts from the same donor (2, 4). However, this concept has been challenged recently by reports of antibody-mediated rejection (AMR) in CLK from the same donor with preexisting DSA (5–9). These case reports and small case series argue against the universal immunoprotection putatively conferred by the liver transplant.
In 2002, the model for end-stage liver disease (MELD) system replaced the United Network for Organ Sharing status classification for the allocation of liver organs (10). In the MELD era, CLK have increased almost threefold, likely, due to the allocation of livers to patients with higher MELD scores, to which serum creatinine and hemodialysis contribute significantly (11). We report findings of our study investigating the impact of allosensitization in CLK on recipient outcomes in a large cohort of transplant recipients using the national Scientific Registry of Transplant Recipients (SRTR) data.
The study population consisted of 2484 CLK recipients with documented panel reactive antibody (PRA) or crossmatch information. Among them, 1882 (76%) had T-cell crossmatch (TXM) information available and 234 (12%) had a positive result. The proportions of sensitized patients based on PRA considering cutoff values of PRA more than or equal to 10%, more than or equal to 20%, and more than or equal to 50%, were 29% (n=692), 24% (n=571), and 15% (n=353), respectively. Overall 747 (30%) of patients had positive TXM, PRA more than or equal to 10%, or both. Among patients with a positive crossmatch, the proportion of PRA=0% was 2.9%, 1% to 10%=3.9%, 11% to 20%=10.2%, and more than 20%=38%. In 2007 to 2008, CLK represented 3.8% of all deceased donor (DD) kidney and 6.3% of DD liver transplants compared with only 1.2% of DD kidney and 2.4% of DD liver transplants in 1995 to 1996. Both the number of CLK performed and the proportions of sensitized recipients increased over the study period (Fig. 1A). As indicated in Table 1, sensitized patients were more likely to be female, African American, with longer pre-transplant dialysis time, kidney re-transplants, and less likely to have diabetes as a primary diagnosis, and receive organs from donors with cardiovascular causes of death. The remaining baseline characteristics were comparable between the study groups.
The proportion of patients receiving tacrolimus and mycophenolate mofetil was not significantly different between sensitized and nonsensitized recipients (65% vs. 62%, respectively, P=0.12). The proportion of patients receiving rabbit antithymocyte globulin was higher among sensitized (19%) versus nonsensitized patients (14%, P=0.003). Steroids were used for maintenance immunosuppression in 80% of the sensitized recipients and in 83% of nonsensitized recipients (P=0.08).
Kidney Allograft Outcomes
Overall kidney graft survival was significantly lower among sensitized patients (log-rank P value=0.015). The 1-, 5-, and 10-year kidney graft survival rates were 75%, 61%, and 43% among sensitized versus 80%, 65%, and 51% among nonsensitized recipients, respectively. This difference in kidney graft survival was evident among patients using criteria of PRA more than 10% alone (P=0.007) but was not statistically different based on patients with only a positive crossmatch (P=0.39). In the multivariable Cox model, the relative hazard of sensitization was independently significant (adjusted hazard ratio=1.16, 95% CI, 1.00–1.36) with overall graft loss.
The proportion of patients with delayed graft function (defined as use of dialysis within the first week posttransplant) was similar between patients with (18%) and without sensitization (17%, P=0.45). The incidence of renal allograft rejection requiring treatment during the initial hospital stay among sensitized recipients was 4.2% vs. 2.9% in nonsensitized recipients, but this difference did not reach statistical significance (P=0.10). There was no difference in death- censored kidney graft loss among the study groups based on allosensitization status.
Based on Kaplan-Meier plots, sensitized patients had significantly reduced patient survival relative to nonsensitized patients (P=0.002, Fig. 1B). These differences in survival translated to estimated half-lives of 10.3 and 7.8 years in the nonsensitized versus sensitized groups, respectively. Patient survival was also significantly reduced among patients who only had PRA more than 10% (P<0.001) but not patients based only on positive crossmatch (P=0.21). In the multivariable Cox model, sensitization had an independent and statistically significantly higher adjusted hazard ratio=1.22 (95% CI, 1.04–1.43) for patient death compared with nonsensitized patients (Table 2). The model also indicated that recipient and donor age, previous kidney and liver transplant recipients, higher MELD scores, and patients transplanted before the MELD era (before 2002) were associated with significantly elevated adjusted hazards for death. Causes of death were relatively comparable between sensitized and nonsensitized recipients (Table 3).
Differences in patient and graft survival were most pronounced in the first year after transplantation. Survival models censored at 1-year follow-up indicated a strong association with time to death (P=0.004) and time to overall kidney graft loss (P=0.012) in the first year. However, neither time to death nor graft loss conditioned on 1-year survival was significantly associated with sensitization beyond the first posttransplant year.
The primary findings of the association between patient survival and sensitization remained unchanged by altering the threshold of PRA. Patient survival was reduced among patients classified by PRA levels more than 20% (P=0.001), more than 50% (P=0.004), and more than 80% (P=0.04) relative to patients with less than the respective thresholds. There was also no significant effect of the method of crossmatch detection between the four primary types (National Institutes of Health method/extended, wash/extended, anti-globulin, and flow cytometry) on patient survival after adjustment for multiple comparisons. Similarly, limiting the analysis to only patients with TXM data (n=1882), the association between sensitization and lower patient and overall kidney graft survival remains significant (P=0.02 and 0.03, respectively).
We additionally examined whether outcomes and differences in survival by sensitization status altered over time. Overall, 1-year survival improved from 77% in 1995 to 1999 to 82% in 2000 to 2004 and 83% in 2005 to 2008. The association between sensitization and lower patient survival remained across eras but varied in magnitude. One-year patient survival was lower and statistically significant between 1995 and 1999 (sensitized 68% vs. nonsensitized 80%, P=0.009), lower but not statistically significant between 2000 and 2004 (80% vs. 83%, P=0.54), and lower and statistically significantly different in most recent years 2005 to 2008 (80% vs. 85%, P=0.03).
The central finding of our study is that patient survival and overall kidney graft survival among CLK recipients were affected by the state of sensitization. In addition, among the growing CLK population in the United States, we note increasing prevalence of sensitization over time. Two previous analyses of SRTR data compared kidney allograft outcomes with the one kidney used in CLK to that of the mate kidney used for kidney transplant alone. One study concluded that the liver neither protects the kidney from rejection nor improves kidney allograft function or survival after CLK (12). The other study reported a lower graft and patient survival rate in the CLK group compared with kidney alone transplantation (KAT) but demonstrated that among human leukocyte antigen (HLA) mismatched and sensitized recipients, rejection free survival was higher among CLK patients (13). Neither cited study was designed a priori to investigate differences in CLK recipient outcomes based on the alloimmunization status. In our study, we specifically investigated whether allosensitization influences recipient outcomes using a reference group of CLK with no evidence of allosensitization.
The number of CLK included in our analysis is more than threefold the number included in the larger of the two referenced studies and we studied survival over a substantially longer follow-up period. These important features have the potential to increase the power to detect small but significant differences in recipient outcomes. We further restricted our analyses to TXM as the clinical relevance of a positive B-cell crossmatch (BXM) remains controversial (14–16). BXM results were not included in the current analysis for a number of reasons. These reasons include the following: (1) BXM information was missing in 55% of the records, (2) in current practice more transplant programs would transplant across a positive BXM but not across TXM, and (3) BXM is more prone to technical issues such as false-positive than TXM.
As has been shown for KAT, sensitized CLK patients were more likely to be women, African Americans, with longer duration of dialysis, and re-transplants (17, 18). Contrarily, one study reported that dialysis duration and gender were not associated with increased sensitization in KAT (19). However, diminished outcomes for sensitized patients in our study were independent of other known risk factors. These results thus suggest that HLA antibodies detected only by PRA are associated with higher incidence of kidney graft loss even in the absence of a positive TXM. This association may be due to antibodies not detected by TXM such as class II or low level class I donor-specific HLA antibodies. Previous reports have associated acute humoral rejection in KAT recipients with pretransplant DSA detected by solid phase assays despite negative Flow XM (20–22). A recent report suggests that XM negative/DSA positive kidney recipients had significantly higher rejection rates at 3 years. They have also shown that more than 50% of positive flow XM recipients were free from rejection, suggesting that a positive XM may not always predict AMR (23). However, this possibility cannot be verified from the registry data and warrants further investigation.
We categorized transplants by year of transplant to minimize possible confounding by advances in immunosuppression, surgical technique, and HLA antibody detection technology over time. In fact, both risk-adjusted patient survival and overall kidney graft survival did improve over time. It is noteworthy that there was a 61% elevated hazard for death in the pre- versus post-MELD era. However, the observed association of sensitization with diminished posttransplant survival was evident in both the pre- and post-MELD eras. The consistency of this lower survival in sensitized patients over time further strengthens the validity of our results.
There were no significant differences in the immunosuppression among sensitized and nonsensitized groups except for higher proportion of patients receiving rabbit antithymocyte globulin among the sensitized patients. However, the exact reasoning and premises underlying selection of immunosuppressive regimens are not well documented in the SRTR data files.
The presented data may not definitively prove that presensitization led to AMR and kidney graft loss due the inherent limitations of registry data and lack of granular patient-level data on rejection episodes, their treatment and the intensity of such treatment. However, the contribution of antibodies as a potential mechanism of kidney allograft loss is plausible in this setting and is consistent with findings of several studies showing that a significant proportion of kidney graft loss is antibody mediated (24–27). Further support of the link between the mechanisms described in these studies and patient survival comes from reports of kidney failure being disadvantageous to patient survival after both renal and nonrenal transplants (28, 29).
The limitations of this study are inherent in the retrospective design. Heterogeneity in the methods used for determination of PRA and crossmatch results in terms of sensitivity and specificity, the inability to discern the effect of class I versus class II antibodies, the strength and donor specificity of HLA antibodies and the incomplete data are important considerations for the interpretation of our findings. However, due to the small number of CLK cases performed in individual centers the registry data analysis provides outcome data of enough CLK to detect associations that are unlikely to be appreciated in a single-center analysis. Although some crossmatch and PRA information is missing in the registry, it is reasonable to assume that missing results are more likely to dilute the effect of sensitization on outcomes, particularly as the observed associations remain consistent regardless of the PRA% threshold used to classify allosensitization status and the sensitivity of the crossmatch methods. In addition, the result of postliver/prekidney transplant crossmatch that is performed in some centers is not currently captured in the registry data. We further noted a paucity of data on sensitization in CLK in the registry. We believe that establishing policies that require complete and accurate reporting of PRA or XM data including postliver/prekidney XM results in the registry would enable future studies to investigate the fate of HLA antibodies and their potential role in graft loss in the setting of CLK. In our experience and as reported previously, many but not all positive preliver flow XM turn negative postliver transplant and this observation may be particularly relevant in an era when increasing proportions of CLK recipients are sensitized (30, 31). High DSA levels may decrease in strength but not turn negative postliver implantation. One may thus speculate that once the liver's capacity to absorb DSA is exceeded, DSA could damage the kidney allograft. This effect may be dependent on factors that are not well quantified such as quality and mass of the liver graft as highlighted by an earlier report of seven highly sensitized patients transplanted with a partial liver and a kidney from the same donor against positive TXM (32). In two of these cases, the TXM remained positive after transplantation, one with a never-functioning renal graft and the other with an early graft failure.
These results provide caution that allosensitization in the setting of CLK is associated with clinically significant reduced recipient survival. Understanding and identifying the characteristics of recipients susceptible to these negative consequences could potentially enable transplant programs to safely maximize CLK. Importantly, the implications of our findings particularly relevant as sensitized CLK recipients are a growing population whose medical care is immensely resource intensive. In view of lower overall kidney graft survival noted in our study and the increased risk of death associated with sensitization in CLK recipients, it also seems prudent to counsel sensitized patients undergoing evaluation for CLK regarding these risks, just as they are typically cautioned about other infrequent complications such as posttransplantation malignancies. This risk perhaps warrants more diligent posttransplant DSA monitoring using protocol timed antibody screening with solid phase assays, and a heightened clinical suspicion index for diagnosis and prompt treatment of AMR in these patients. In addition, administration of more potent immunosuppression at the upper therapeutic ranges when posttransplant crossmatches remain positive may also be warranted. In a recent study, a sensitized CLK recipient whose flow TXM remained positive postliver implantation, developed a severe AMR episode that responded only to a Bortezomib-based rescue treatment (33).
Although the observed increased risk in sensitized CLK noted in our study does not seem to be large enough to discourage the decision to pursue a life saving CLK, our findings do suggest that sensitization should be studied systematically as a factor for risk adjustment; especially with regard to specific causes of patient death and kidney graft loss. Future studies should focus on donor specificity of HLA antibodies, antibody levels and kinetics in relation to liver implantation, immunosuppressive management, creatinine slopes, and changes in renal allograft histology. Details of AMR episodes and interventions directed at reducing antibody levels in these patients will also be critical to delineating specific effects of DSA and sensitization on outcome in this complex clinical setting.
In conclusion, overall kidney allograft and patient survival are diminished significantly in sensitized CLK recipients. These clinically relevant findings need further investigation in well-designed prospective studies that can elucidate immunologic and nonimmunologic causative mechanisms.
MATERIALS AND METHODS
In this study, we used data from the national SRTR registry. The SRTR system includes data on all donor, wait-listed candidates, and transplant recipients in the United States, submitted by the members of the Organ Procurement and Transplantation Network, and has been described elsewhere (34). The study was approved by the Cleveland Clinic Institutional Review Board.
We evaluated adult (aged 18 years and older) transplant recipients who received CLK DD transplants from 1995 to 2008. We identified CLK by merging solitary liver and solitary kidney transplant files to construct an analysis file of patients with the same transplant dates or dates within 2 days for the transplant procedures. The primary explanatory variable of the models was an indication of a positive TXM or PRA more than or equal to 10% at the time of transplantation and are referred to as sensitized CLK recipients for the purposes of this analysis. All other CLK recipients including patients with PRA less than 10% and negative or missing TXM are referred to as nonsensitized CLK recipients. Patients missing both PRA levels and TXM information were excluded (n=593; 19%). The study inclusion and distribution of patients classified as sensitized are displayed in Figure 2.
Methods used to determine a positive TXM included Flow cytometry (n=698; 37%), Anti-globulin (n=878; 47%), and Complement Dependent Cytotoxicity (CDC)/other (n=306; 16%). BXM information was missing in 55% of the patients and was not considered in our analysis.
Definitions and Statistical Analysis
We used Kaplan-Meier and Cox proportional hazard models to assess the association of patient sensitization with the primary outcomes of time to patient death and kidney graft loss defined as return to maintenance dialysis, re-transplantation, or death. Covariates in the survival models included donor and recipient age, gender, body mass index and race, primary kidney and liver diagnosis, Hepatitis C status, MELD score, HLA-mismatching, and donor cause of death. As MELD score was not available in the database before implementation into policy in 2002, a level of “pre-MELD” was used for this set of patients representing transplants before the use of MELD scores for allocation. Assumptions for the proportional hazard models were verified by visual inspection of the complementary log-log survival plots and interaction terms with time for continuous covariates. Acute rejection episodes were defined as indications for treatment of acute rejection as documented on the United Network for Organ Sharing registry follow-up forms.
To assess the impact of the sensitization on early kidney allograft outcomes, we compared the incidence of rejection requiring treatment during the initial hospital stay among sensitized versus nonsensitized CLK recipients. We also quantified long-term survival based on sensitization status by estimating patients' half-life. Patient half-life was estimated by calculating the time to 50% patient survival from the day of transplantation derived from the Kaplan-Meier plots. All reported P values are two-sided, and a P value of less than 0.05 was considered statistically significant. All analyses were conducted in SAS version 9.2 (Cary, NC).
1. Margreiter R, Kramar R, Huber C, et al. Combined liver and kidney transplantation
1984; 1: 1077.
2. Fung J, Makowka L, Tzakis A, et al. Combined liver-kidney transplantation
: Analysis of patients with preformed lymphocytotoxic antibodies. Transplant Proc
1988; 20: 88.
3. Kissmeyer-Nielsen F, Olsen S, Petersen VP, et al. Hyperacute rejection of kidney allografts, associated with pre-existing humoral antibodies against donor cells. Lancet
1966; 2: 662.
4. Kamada N, Davies HS, Roser B. Reversal of transplantation immunity by liver grafting. Nature
1981; 292: 840.
5. Starzl TE, Demetris AJ, Todo S, et al. Evidence for hyperacute rejection of human liver grafts: The case of the canary kidneys. Clinical Transplantation
1989; 3: 37.
6. Eid A, Moore SB, Wiesner RH, et al. Evidence that the liver does not always protect the kidney from hyperacute rejection in combined liver-kidney transplantation
across a positive lymphocyte crossmatch. Transplantation
1990; 50: 331.
7. Hadaya K, Ferrari-Lacraz S, Giostra E, et al. Humoral and cellular rejection after combined liver-kidney transplantation
in low immunologic risk recipients. Transpl Int
2009; 22: 242.
8. Reichman TW, Marino SR, Milner J, et al. Acute humoral rejection in an ABO compatible combined liver-kidney transplant–the kidney is not always protected. Am J Transplant
2009; 9: 1957.
9. Saidman SL, Duquesnoy RJ, Demetris AJ, et al. Combined liver-kidney transplantation
and the effect of preformed lymphocytotoxic antibodies. Transpl Immunol
1994; 2: 61.
10. Freeman RB Jr, Wiesner RH, Harper A, et al. The new liver allocation system: Moving toward evidence-based transplantation policy. Liver Transpl
2002; 8: 851.
11. Dube GK, Cohen DJ. Simultaneous liver and kidney transplantation
. Curr Opin Nephrol Hypertens
2007; 16: 547.
12. Katznelson S, Cecka JM. The liver neither protects the kidney from rejection nor improves kidney graft survival after combined liver and kidney transplantation
from the same donor. Transplantation
1996; 61: 1403.
13. Fong TL, Bunnapradist S, Jordan SC, et al. Analysis of the United Network for Organ Sharing database comparing renal allografts and patient survival
in combined liver-kidney transplantation
with the contralateral allografts in kidney alone or kidney-pancreas transplantation. Transplantation
2003; 76: 348.
14. Bryan CF, Aeder MI, Helling TS, et al. Clinical relevance of a positive B-cell crossmatch on renal transplantation: A multi-transplant center evaluation. Transplant Proc
1993; 25: 247.
15. Mahoney RJ, Taranto S, Edwards E. B-cell crossmatching and kidney allograft outcome in 9031 United States transplant recipients. Hum Immunol
2002; 63: 324.
16. Eng HS, Bennett G, Tsiopelas E, et al. Anti-HLA donor-specific antibodies detected in positive B-cell crossmatches by Luminex predict late graft loss. Am J Transplant
2008; 8: 2335.
17. Rosenberg JC, Beyersdorf TM, Derbyshire N, et al. Retrieval and allocation of organs for transplantation: The Michigan experience. Clin Transpl
18. Katznelson S, Bhaduri S, Cecka JM. Clinical aspects of sensitization. Clin Transpl
19. Pour-Reza-Gholi F, Daneshvar S, Nafar M, et al. Potential risk factors for hypersensitization reflected by panel-reactive antibodies in dialysis patients. Transplant Proc
2005; 37: 2936.
20. Ishida H, Tanabe K, Furusawa M, et al. Evaluation of flow cytometric panel reactive antibody in renal transplant recipients—Examination of 238 cases of renal transplantation. Transpl Int
2005; 18: 163.
21. Miura M, Kubota KC, Itoh T, et al. Positive pre-transplant flow-panel reactive antibody detected after accelerated acute rejection with negative pre-transplant flow crossmatch. Clin Transplant
2006; 20(suppl 15): 33.
22. Bielmann D, Honger G, Lutz D, et al. Pretransplant risk assessment in renal allograft recipients using virtual crossmatching. Am J Transplant
2007; 7: 626.
23. Dunn TB, Ozturk OG, Noreen H, et al. Impact of flow XM results on kidney transplant outcomes for recipients with donor specific antibody (DSA) [abstract]. Am J Transplant
2010; 10 (suppl 4): 236.
24. El-Zoghby ZM, Stegall MD, Lager DJ, et al. Identifying specific causes of kidney allograft loss. Am J Transplant
2009; 9: 527.
25. Einecke G, Sis B, Reeve J, et al. Antibody-mediated microcirculation injury is the major cause of late kidney transplant failure. Am J Transplant
2009; 9: 2520.
26. Gaston RS, Cecka JM, Kasiske BL, et al. Evidence for antibody-mediated injury as a major determinant of late kidney allograft failure. Transplantation
2010; 90: 68.
27. Cai J, Terasaki PI, Anderson N, et al. Intact HLA not beta2m-free heavy chain-specific HLA class I antibodies are predictive of graft failure. Transplantation
2009; 88: 226.
28. Kaplan B, Meier-Kriesche HU. Death after graft loss: An important late study endpoint in kidney transplantation
. Am J Transplant
2002; 2: 970.
29. Ojo AO, Held PJ, Port FK, et al. Chronic renal failure after transplantation of a nonrenal organ. N Engl J Med
2003; 349: 931.
30. Askar M, Smith G, Klingman L, et al. Liver transplantation
significantly decreases donor HLA specific antibodies in some but not all sensitized recipients of combined liver/kidney transplants [abstract]. Int J Immunogenet
2010; 37: 417.
31. Manez R, Kelly RH, Kobayashi M, et al. Immunoglobulin G lymphocytotoxic antibodies in clinical liver transplantation
: Studies toward further defining their significance. Hepatology
1995; 21: 1345.
32. Olausson M, Mjornstedt L, Norden G, et al. Successful combined partial auxiliary liver and kidney transplantation
in highly sensitized cross-match positive recipients. Am J Transplant
2007; 7: 130.
33. Flechner SM, Fatica R, Askar M, et al. The role of proteasome inhibition with bortezomib in the treatment of antibody-mediated rejection after kidney-only or kidney-combined organ transplantation. Transplantation
2010; 90: 1486.
34. Dickinson DM, Arrington CJ, Fant G, et al. SRTR program-specific reports on outcomes: A guide for the new reader. Am J Transplant
2008; 8: 1012.