- Top of page
- Materials and Methods
- Authors' Contributions
In 2006, Melcher et al  described a case of antibody-mediated rejection (AMR) in the pancreas after simultaneous pancreas–kidney (SPK) transplantation where pancreatic graft dysfunction was accompanied by positive C4d staining on biopsy and positive donor-specific antibody (DSA). This was successfully treated with plasmapheresis, intravenous immune globulin (IVIG) and rituximab. Since then, further reports and case series have confirmed the existence of pancreatic AMR as its own entity [2-6].
Following kidney or heart transplantation, AMR is well recognized and diagnostic criteria are established [7, 8]. However, AMR following pancreas transplantation is a more newly defined entity. While recently proposed key criteria of pancreatic graft AMR include capillaritis, interacinar capillary C4d staining, graft dysfunction and detectable DSA [3, 9, 10], the reported experience using these criteria is based on small case series from a few centers [2, 3, 6, 11]. Thus, many important questions related to AMR of the pancreatic allograft remain unanswered.
Specifically, risk factors for pancreatic graft AMR have not been described and outcomes have not been systematically evaluated. We report a large single-center experience with pancreas biopsies and the diagnosis of AMR after pancreas transplantation.
Furthermore, although the presence of DSA was required in the initial Banff criteria , it is no longer an absolute requirement for a diagnosis “consistent with AMR” in the new Banff criteria . To further define the impact of DSA in the diagnosis of pancreatic allograft AMR, we analyzed DSA data and correlated DSA findings with C4d staining in pancreatic allograft biopsies. Herein, we report the frequency of the diagnosis of AMR in the pancreatic allograft in a typical pancreas transplant population while describing risk factors for AMR and functional outcomes.
Materials and Methods
- Top of page
- Materials and Methods
- Authors' Contributions
Since August 2006, patients undergoing pancreas allograft biopsies had prospective C4d staining performed and serum DSA levels determined by Luminex® assay. We conducted a retrospective chart review of patients who received an SPK transplant or solitary pancreas (PAN) transplant between August 1, 2006 and December 31, 2009. PAN transplants included pancreas after kidney transplant, pancreas transplant alone or pancreas after SPK transplants. During the study period, 159 patients underwent 162 pancreas transplants (Figures 1 and 2A). Our end date was chosen with the goal of having greater than 1-year follow-up data on all patients. The institutional review board at the University of Wisconsin-Madison School of Medicine and Public Health approved this study.
Figure 1. Outline of patients, transplants and biopsies included in this study. Some transplants were biopsied more than once. Retransplants were defined as any subsequent transplant (e.g. a pancreas after kidney transplant would be considered a retransplant).
Download figure to PowerPoint
Figure 2. (A) Distribution of the types of pancreas transplant. (B) Number of pancreas transplant biopsies per pancreas allograft. (C) Indications for pancreas transplant biopsy. (D) Pancreas allograft biopsy method. SPK, simultaneous pancreas–kidney; DSA, donor-specific antibody; CT, computed tomography.
Download figure to PowerPoint
Donor and recipient demographics are listed in Table 1. All pancreas transplants were performed using systemic venous and enteric exocrine drainage. All transplants had a negative flow crossmatch. Induction therapy was either basiliximab (Simulect; Novartis, East Hanover, NJ), alemtuzumab (Campath; Genzyme, Cambridge, MA) or antithymocyte globulin (ATG; Thymoglobulin; Genzyme), depending on the era and type of transplant. Maintenance immunosuppression included prednisone (30 mg/day at discharge, tapered to 10 mg/day by month 3 posttransplant), mycophenolate sodium 3 g/day (Myfortic; Novartis) and tacrolimus (Prograf; Astellas, Deerfield, IL), with goal levels of 8–11 ng/mL during the first year, then tapering to 6–10 ng/mL thereafter.
Table 1. Demographics of 162 pancreas transplants in 159 recipients
|Age||41 ± 9 years||Age||32 ± 12 years|
|CMV negative||62%||CMV negative||50%|
|BMI||25 ± 3.5||SCD/DBD||83%|
|Duration||27 ± 9 years||SCD/DCD||15%|
|Waitlist||167 ± 209 days||ECD||2%|
|Peak PRA||4 ± 11%||CIT||13 ± 4 h|
During the study, we performed 94 pancreas transplant biopsies in 49 pancreas allografts for various indications (Figures 1 and 2B and C). One hundred thirteen transplants were never biopsied. Of these, 28 were treated empirically for rejection and these were excluded from the analyses. Most biopsies were performed percutaneously with real-time ultrasound guidance using a needle guide and obtaining at least one full 18 gauge, 22 mm long core (Figure 2D). Tissue cores were prepared as previously described . Of note, C4d staining was done by immunohistochemistry on paraffin sections. Biopsies were scored for rejection based on the Banff criteria for pancreatic allograft rejection [9-11]. Serial biopsies performed on the same patient within 30 days were considered to be a single episode of rejection and the highest pathology grade of the multiple biopsies was assigned as the representative biopsy.
Our current algorithm for treatment of pancreas transplant rejection is to use steroids for grade I acute cellular rejection (ACR), and steroids and thymoglobulin (1.5 mg/kg/day for 5–7 days) for grade II and III ACR. We treat early (<90 days) pancreas transplant AMR with steroids, plasmapheresis and IVIG (100 mg/kg/week, maximum 7 g/week), and late (>90 days) AMR with steroids and a 4-week course of IVIG, with the possible addition of rituximab in either setting. The treatment regimen for pancreas allograft AMR is similar to that for kidney allograft AMR at our institution .
Diagnosis of pancreatic allograft rejection
Pancreas allograft rejection was diagnosed by pancreatic graft biopsy and episodes were categorized by biopsy findings as ACR, AMR or mixed rejection. ACR was established if histology showed grade I or higher rejection based on Banff criteria . AMR was defined as (i) graft dysfunction, (ii) focal (5–50%) or diffuse (>50%) linear C4d deposition along interacinar capillaries and (iii) detectable DSA , which was defined as normalized maximum mean fluorescence intensity (MFI) (i.e. actual MFI value minus cutoff MFI value of the anti-HLA antibody allele with the highest normalized MFI value) greater than 500. DSA was determined by Luminex® single antigen bead assay (One Lambda, Inc., Canoga Park, CA). At our institution, a single DSA of 2500 MFI is expected to result in a positive crossmatch. Mixed rejection was defined as ACR and AMR occurring simultaneously. The “overall rejection” group included all biopsy-proven ACR, AMR and mixed rejection episodes, which we termed “Any Pancreas BPAR.” Empirically treated rejection episodes were excluded from the analyses. Rejection rates were reported in a time-to-event model. Fasting glucose, amylase and lipase serum levels were collected and analyzed at the time of biopsy and compared between rejection groups using Student's t-test.
Risk factor analysis
Univariate and multivariate analyses were performed to identify potential risk factors for rejection. Risk factors included in the analyses are listed in Table 2. Endpoints for risk factor analyses were: (i) any biopsy-proven acute pancreas rejection, and (ii) AMR specifically. Impactful risk factors were reported as hazard ratios with 95% confidence intervals. Kaplan–Meier curves were generated to estimate 1-year rejection rates.
Table 2. Risk factors for rejection of the pancreas
|Any pancreas BPAR||Pancreas AMR||Any pancreas BPAR||Pancreas AMR|
|HR||95% CI||p-Value||HR||95% CI||p-Value||HR||95% CI||p-Value||HR||95% CI||p-Value|
|Recipient age (every 10 years older)||0.604||0.404–0.904||0.014||0.557||0.311–0.997||0.049||0.713||0.473–1.073||0.105||0.662||0.363–1.209||0.179|
|Donor age (every 10 years older)||0.943||0.715–1.244||0.677||1.057||0.713–1.567||0.784||–||–||–||–||–||–|
|Solitary pancreas transplant||2.555||1.520–5.810||0.001||2.339||0.890–6.146||0.085||–||–||–||–||–||–|
|Primary solitary pancreas transplant||4.684||2.048–10.714||0.000||4.367||1.231–15.492||0.023||4.422||1.846–10.592||0.001||4.362||1.144–16.630||0.310|
|Nonprimary solitary pancreas transplant||2.791||1.081–7.208||0.034||2.960||0.739–11.856||0.125||2.222||0.813–6.077||0.120||1.835||0.418–8.044||0.421|
|Recipient CMV positive||0.609||0.292–1.270||0.186||0.461||0.150–1.419||0.177||–||–||–||–||–||–|
|Donor CMV positive||0.921||0.474–1.792||0.809||0.753||0.286–1.980||0.565||–||–||–||–||–||–|
|CMV negative to negative recipient||2.396||0.675–8.502||0.176||4.356||0.534–35.531||0.169||–||–||–||–||–||–|
|CMV negative to positive recipient||2.104||0.544–8.140||0.281||2.809||0.292–27.024||0.371||–||–||–||–||–||–|
|CMV positive to negative recipient||2.813||0.801–9.883||0.107||4.024||0.483–33.499||0.198||–||–||–||–||–||–|
|Higher recipient BMI||0.997||0.908–1.096||0.957||1.056||0.927–1.204||0.411||–||–||–||–||–||–|
|Longer duration of disease pretransplant||0.995||0.959–1.033||0.805||0.978||0.927–1.033||0.432||–||–||–||–||–||–|
|More days on the waitinglist||0.999||0.997–1.001||0.351||0.998||0.995–1.002||0.347||–||–||–||–||–||–|
|Simulect induction (vs. other induction)||1.742||0.851–3.567||0.129||1.329||0.479–3.693||0.585||–||–||–||–||–||–|
|Total HLA mismatch > 4 antigens||1.070||0.552–2.077||0.840||0.869||0.335–2.253||0.773||–||–||–||–||–||–|
|Higher peak PRA||0.982||0.936–1.029||0.443||0.998||0.951–1.046||0.920||–||–||–||–||–||–|
|Standard criteria donor (vs. DCD)||0.868||0.360–2.091||0.752||0.819||0.235–2.851||0.753||–||–||–||–||–||–|
|Longer pancreas cold ischemic time||1.001||0.999–1.002||0.281||1.001||0.999–1.003||0.437||–||–||–||–||–||–|
The correlation between positive C4d staining and positive DSA
DSA data were collected on most patients within 1 month of the pancreas transplant biopsy as per our clinical protocol. During the study period, 79 biopsies had data on both C4d staining and HLA Class I and/or Class II serum levels of DSA. For this analysis, the logarithmic value of the antibody with the maximum MFI value of each Class I and Class II DSA above cutoff was used to minimize differences in standard deviation. We defined four groups by grade of C4d staining: C4d negative, C4d <5% (minimal), C4d 5–50% (focal) and C4d >50% (diffuse). We analyzed these groups with respect to their serum levels of DSA using analysis of variance (ANOVA) and Fisher's protected least significant difference test. Data are reported as mean ± standard deviation (SD). p-Values <0.05 were considered significant.
Receiver–operator characteristic (ROC) graphs were generated by plotting the relationship between the sensitivity against 1-specificity at various maximum MFI values and calculating the resulting area under the curve (AUC). The gold standard was defined as positive (>5%) C4d staining and the variable was a positive maximum MFI of either Class I or Class II or both. An AUC >0.9 is considered a strong predictor, whereas an AUC of 0.5 indicates random association between the test measure and the gold standard, and falls along the line of identity.
Definitions of positive DSA
Seventy-one pancreas transplant biopsies had data on both Class I and Class II DSA. The percentage of C4d staining was correlated to different definitions of “positive DSA” to determine the most meaningful way of interpreting DSA data. All DSA data were normalized by subtracting the cutoff value from the allele-specific MFI value. Thereafter, DSA was examined in the following ways: (i) “Summed MFI” was calculated by adding all positive MFI values together. Analysis was carried out considering only summed anti-donor Class I antibodies, or only summed anti-donor Class II antibodies or both anti-donor Class I and Class II antibodies added together. (ii) Maximum Class I or Class II MFI was defined as the highest MFI for each anti-donor Class I or Class II antibody, or the higher of Class I or II antibody values, and using this number only if it was greater than zero.
The extent of C4d staining was stratified into four groups as described earlier. Groups were analyzed with respect to each definition of DSA using ANOVA and Fisher's protected least significant difference test and Fisher's exact test. Results are given as mean ± standard deviation. p-Values <0.05 were considered significant.
To evaluate postrejection outcomes, we more specifically evaluated 35 pancreas transplant biopsies, which were the first pancreas biopsies performed that demonstrated evidence of rejection after any given pancreas transplant. Biopsies that did not show rejection and biopsies done after a prior biopsy had already confirmed rejection were excluded for this particular analysis. AMR was defined as graft dysfunction in the presence of C4d staining >5% and positive DSA (>500 Class I or Class II MFI). As precise cutoffs for DSA are not defined, for the purpose of comparison, positive DSA was defined as at least one antibody with an MFI > 500 above Luminex® control. Using Kaplan–Meier analyses, we compared outcomes of early (within 90 days posttransplant) versus late (90 days or more after transplant) AMR. Similarly, we compared outcomes by rejection type: pure AMR versus pure ACR, and mixed (AMR/ACR) rejection.
- Top of page
- Materials and Methods
- Authors' Contributions
Pancreatic graft AMR occurs at a meaningful frequency in patients following all types of pancreas transplantation. Overall in this cohort we found 21% of pancreas transplant recipients were treated for any type of BPAR within the first posttransplant year and the incidence of AMR of the pancreatic allograft is 10% when using a definition of graft dysfunction, >5% C4d staining of interacinar capillaries, and Class I or II MFI >500. To our knowledge, no prior literature reports prospectively examine the incidence of AMR after pancreas transplantation. In Rangel et al.'s  experience, 43% of biopsied pancreatic allografts showed positive staining for C4d. While this could be regarded as a rough estimate of the incidence of AMR in this population, DSA data were lacking and not all patients underwent biopsy , which would lead to an overestimation of the true incidence of AMR. The 10% incidence of AMR in our cohort was found in the setting of a significant fraction of retransplants (18% of the total pancreas transplant population). Other papers have described a comparable incidence of AMR of the kidney allograft after SPK transplantation between 13% and 43%, depending on the level of pretransplant recipient sensitization [2, 12, 13]. Our rate of pancreatic graft AMR is also comparable to the incidence of AMR reported in the kidney transplant population, which ranges from 7.7% to 41% with the risk being greater if recipients are sensitized [14-17]. This comparison suggests that pancreas grafts are not more predisposed to AMR than kidney transplants, yet to truly answer this question a larger cohort is necessary. Nonetheless, given the frequency of AMR after pancreas transplantation, efforts should be made to establish or eliminate this diagnosis in the setting of graft dysfunction in order to more specifically guide immunosuppressive therapy.
Given the sensitized state of the retransplant group it is not unexpected that risk factors for AMR include repeat SPK transplantation, and pancreas transplantation involving a donor–recipient race mismatch. Primary PAN transplantation is also a risk factor for both AMR and any type of BPAR in this series. While female gender was reported a risk factor for AMR of the kidney after SPK transplantation , this was not found to be a risk factor for pancreatic graft AMR here. Donor–recipient race mismatch as a risk factor for pancreas AMR was a somewhat surprising finding. The reason(s) for this observation are unknown but may be due to HLA repertoire disparities. Based on these findings, we believe special consideration should be given to recipients of PAN transplants or pancreas retransplants in terms of altering induction immunosuppression, considering increased maintenance immunosuppression and increased surveillance during the early posttransplant course.
Nationwide, a lack of standardization of DSA measurements remains . While evidence suggests that elevated de novo DSA is a harbinger of rejection and poor outcomes in kidney transplantation, it is not yet clear whether changes in DSA (or a newly positive DSA >500 MFI above cutoff) in the setting of allograft dysfunction without biopsy confirmation can be used to monitor for AMR in the setting of pancreas transplantation . The low c-indices found in the ROC analysis of DSA and C4d staining indicate that elevated DSA alone is neither sensitive nor specific for pancreatic graft AMR, thus justifying the current Banff criteria . We believe it is reasonable to routinely monitor posttransplant DSA and to perform a pancreas allograft biopsy if a significant increase in DSA occurs. However, we do not treat increases in DSA without biopsy confirmation. Interpretation and management of minimally and focally C4d positive biopsies requires clinical correlation.
Furthermore, the best method of quantifying DSA intensity has not been carefully examined. Based on our results, maximum Class I DSA correlates better with C4d staining than maximum Class II DSA. Specifically, very high maximum Class I DSA values correlate with diffusely C4d positive biopsies, which is consistent with other reports [19, 20]. Interestingly, both mixed rejection and diffusely positive C4d biopsies are often found later posttransplant than minimally or focally positive C4d staining biopsies. This suggests that a rise in Class I DSA (to >500 MFI above cutoff at our institution) and a finding of minimal or focal C4d staining on an early biopsy may represent the beginning of AMR. Thus consideration should be given to early treatment, possibly even in the absence of other clinical signs of rejection. Interestingly, summed Class I MFI results in the strongest correlation between DSA and C4d staining, suggesting that cumulative alloantibody load is important.
For the present study, positive DSA was defined as anti-HLA Class I or II DSA MFI >500 above cutoff. With this threshold the incidence of AMR after pancreas transplantation was 10%. A previous study showed C4d staining in patients with underlying pancreatic allograft dysfunction was strongest in patients whose MFI was either >500 above cutoff for Class I DSA or >500–1000 above cutoff for Class II DSA . The present study confirms prior findings that an MFI >500 may serve as a reasonable definition for AMR. This cutoff is further corroborated by observations in a number of studies that de novo DSA between >300 and >1000 MFI as determined by single antigen bead assays in kidney transplantation correlates with poor graft survival outcomes . Whereas in kidney transplantation, the presence of Class II DSA portends a worse prognosis, we cannot say this with certainty in pancreas transplantation. Our findings demonstrate that elevated Class I DSA correlates more strongly with C4d staining than Class II DSA, but whether this impacts the outcome of the pancreatic allograft is as of yet unknown. To determine such a causal relationship would require a larger number of patients and longer-term follow-up. Nonetheless, we feel rising or de novo Class II DSA posttransplant is important as many patients with biopsy-proven AMR or mixed rejection exhibited increasing Class II DSAs, especially DQ antibodies.
As ultrasound-guided percutaneous pancreas transplant biopsy is safe and easily accessible at our institution, we infrequently treat pancreatic allograft rejection empirically. Empiric treatment of enzyme elevations without a biopsy is considered within the first 3 days posttransplantation, in the setting of anticoagulation, or if there is no safe window for biopsy on ultrasound or computed tomography imaging and no other possible cause is identified on abdominal imaging. Even then, we will sometimes perform laparoscopic or open pancreas transplant biopsies. Biopsy and a full histopathologic analysis with C4d staining provide the opportunity to institute focused and specific therapy for each type of rejection (i.e. AMR, ACR or mixed) in any given patient.
Whether mixed pancreas rejection has worse long-term outcomes than ACR or AMR remains uncertain at this time. In evaluating the outcomes after 35 initial episodes of pancreas transplant rejection, AMR did not carry an obviously worse prognosis than ACR; and mixed rejection also did not result in significantly worse allograft survival during short-term follow-up. Given previous reports of worse graft outcomes after AMR and mixed rejection in both the kidney and pancreas transplant literature, this is a potentially surprising finding [2, 20]. However, the result can be interpreted in several ways: First, the number of subjects in each group was low, and follow-up was limited to just over 1 year. Thus, differences in outcomes may be found once follow-up increases or as more subjects are available for study. Second, depending on clinical findings, biopsy findings and DSA, treatment for rejection was quite variable (data not shown), which may explain our good short-term results. As the experience with rejection after pancreas transplantation increases, it will be interesting to tease out how treatment affects outcomes after different types of pancreas transplant rejection. Furthermore, given a 19% incidence of graft failure at 2 years and 27% incidence of repeat rejection episodes in this study, close follow-up is warranted and consideration should be given to follow-up or surveillance biopsies. Such biopsies may become especially important if there is incomplete resolution of the presenting findings, such as persistent elevations in amylase or lipase.
Prior to the recognition of AMR, pancreas allograft rejection was most commonly treated with a steroid pulse taper, and ATG was added if rejection was considered severe on biopsy or if there was no clinical improvement with steroids alone. With the availability of C4d staining and its proven correlation with AMR , as well as with the established and updated Banff criteria for pancreatic allograft rejection [9, 10], pancreas transplant biopsy has become the key diagnostic tool in the management of pancreas transplant recipients.
In conclusion, AMR of the pancreatic allograft occurs in 1 out of 10 patients during the first year posttransplant. Risk factors for the development of pancreatic AMR are not modifiable; thus, adjusting immunosuppression and careful monitoring are necessary to minimize the incidence of AMR. Patient survival after rejection is 100%, and 1-year postrejection allograft survival is above 80%. Thus, outcomes after pancreas allograft rejection, including AMR, are excellent with appropriate treatment.