Acute Pancreas Allograft Rejection Is Associated With Increased Risk of Graft Failure in Pancreas Transplantation

American Journal of Transplantation 2013; 13: 1019 1025 Wiley Periodicals Inc. C Copyright 2013 The American Society of Transplantation and the American Society of Transplant Surgeons doi: 10.1111/ajt.12167 Acute Pancreas Allograft Rejection Is Associated With Increased Risk of Graft Failure in Pancreas Transplantation M. Dong a,b, A. K. Parsaik a,w.kremers c, A. Sun a,d, P. Dean e,m.prieto e,f.g.cosio f, M. J. Gandhi g, L. Zhang h,t.c.smyrk h, M. Stegall D. Stegall e and e and Y. C. Y. Kudva C. Kudva a, a, * a Division of Endocrinology, Diabetes, Nutrition, and Metabolism, Department of Internal Medicine, Mayo Clinic, Rochester, MN b Department of Endocrinology and Metabolism, Qilu Hospital of Shandong University, Jinan, Shandong, P. R. China c Division of Biomedical Statistics and Informatics, Department of Health Sciences Research, Mayo Clinic, Rochester, MN d Department of Endocrinology and Metabolism, Zibo First People s Hospital, Zibo, Shandong, P. R. China e Division of Transplantation Surgery, Department of Surgery, f Division of Nephrology and Hypertension, Department of Internal Medicine, g Division of Transfusion Medicine, Department of Laboratory Medicine and Pathology and h Division of Anatomic Pathology, Department of Laboratory Medicine and Pathology, Mayo Clinic, Rochester, MN Corresponding author: Yogish C. Kudva, kudva.yogish@mayo.edu The effect of acute allograft rejection (AR) on long-term pancreas allograft function is unclear. We retrospectively studied 227 consecutive pancreas transplants performed at our institution between January 1, 998 and December 31, 2009 including: 56 simultaneous pancreas and kidney (SPK), 69 pancreas transplantation alone (PTA); and 102 pancreas after kidney (PAK) transplants. With a median follow-up of 6.1 (IQR 3 9) years, 57 patients developed 79 episodes of AR, and 19 experienced more than one episode. The cumulative incidence for AR was 14.7%, 19.7%, 26.6% and 29.1% at 1, 2, 5 and 10 years. PTA transplant (hazards ratio [HR] = 2.28, p = 0.001) and donor age (per 10 years) (HR = 1.34, p = 0.006) were associated with higher risk for AR. The first AR episode after 3 months post PT was associated with increased risk for complete loss (CL) (HR 3.79, p < 0.001), and the first AR episode occurring during 3- to 12-month and 12- to 24-month periods after PT were associated with significantly increased risk for at least partial loss (PL) (HR 2.84, p = 0.014; and HR 6.25, p < 0.001, respectively). We conclude that AR is associated with increased risk for CL and at least PL. The time that the first AR is observed may influence subsequent graft failure. Key words: Acute rejection, complete graft loss, pancreas transplantation, partial graft loss Abbreviations: AR, acute rejection; ATG, antithymocyte globulin; CL, complete loss; CMV, cytomegalovirus; DSA, donor-specific antibody; FF, full function; HR, hazards ratio; MFI, mean fluorescence intensity; PAK, pancreas after kidney transplantation; PL, partial loss; PTA, pancreas transplantation alone; PRA, panel reactive antibody; PT, pancreas transplantation; SPK, simultaneous pancreas kidney transplantation; T1D, type 1 diabetes mellitus; UMD, University of Maryland. Received 10 July 2012, revised 18 December 2012 and accepted for publication 26 December 2012 Introduction Pancreas transplantation (PT) has been performed for more than 40 years, with continual improvement in both patient and graft survival (1,2). A successful PT can eliminate acute metabolic complications and stabilize some of the longterm complications of diabetes (3). The 1-year graft survival rates have increased to 86%, 79% and 80% for simultaneous pancreas and kidney (SPK), pancreas after kidney (PAK) and pancreas transplantation alone (PTA) transplants, respectively, due to improved surgical techniques, more potent immunosuppression, the use of chemoprophylaxis for the prevention of infections and careful monitoring of allograft functional status (4). Allograft rejection is still one of the leading causes of allograft loss, and pancreas loss due to rejection may subsequently influence patient survival (5,6). Reported rejection rates vary depending on the immunosuppression protocol used (3). According to data reported to the United Network for Organ Sharing and the International Pancreas Transplant Registry, acute rejection (AR) as a cause of graft loss was most frequent at 7 12 months posttransplant and decreased after the first year (7). The rejection loss rates at 1 year had declined to 8% in the solitary (PAK and PTA) categories and 2% in the SPK category in 2002/2003 (1). However, the incidence of AR was not reported in that study, and the effect of AR on long-term graft function has not been well documented in the literature, especially in PTA and PAK cohorts. 1019

Dong et al. An association between late AR and reduced long-term graft survival has been reported in kidney transplantation recipients in the literature but has been reported to a limited extent after pancreas transplantation thus far (8 11). Early studies showed late AR, defined as a first episode of AR occurring after 1-year post PT, had more adverse consequences in PTA and PAK recipients when compared to SPK recipients (12,13). To our knowledge, the impact of late AR on long-term pancreas allograft function has not been well described. We reviewed our experience of PT at the Mayo Clinic from 1998 through 2009. The majority of the subjects (75%) had a solitary pancreas transplant (PAK and PTA). The main purpose of this study is to determine the effect of treated biopsy-proved AR on pancreas allograft function and survival. Materials and Methods The study was approved by the Mayo Clinic Institutional Review Board and all included patients provided informed consent. Study design and data collection We performed a retrospective review of all adult patients ( 18 years) who received a first PT at Mayo Clinic, Rochester, Minnesota, from January 1, 1998 to December 31, 2009. Reasons for exclusion included loss of the pancreas graft within 2 weeks of transplantation (primarily due to early graft thrombosis) or lack of research authorization. Two hundred and twentyseven recipients were eligible for final analysis. Data were collected using medical records, including basic demographics, type of PT, CMV recipient and donor pretransplant serological status, HLA-mismatch, PRA, type of exocrine drainage, induction and maintenance immunosuppression, biopsy-confirmed AR, treatment for AR and pancreas allograft and patient survival outcomes. Prior to 2006, PRA was determined by a CDC-AHG assay using 56 or 60 well commercial T-lymphocyte frozen cell tray (Gentrak Inc., Liberty, NC, USA) per the manufacturer s instructions. A positive reaction was defined as lysis of >50% of the cells. After 2006, anti-hla antibodies were identified using the LABScreen Single Antigen Bead assay (One Lambda Inc., Canoga Park, CA, USA) (14). Mean fluorescence intensity (MFI) value >5000 was considered positive. Low-resolution Class I and Class II typing were performed either by molecular or serologic methods. For typing done by molecular method serologic equivalents were reported. Initial and maintenance immunosuppression treatment The immunosuppressive protocols used in these patients have been described previously (15,16). In brief, all recipients received induction immunosuppressive therapy by using rabbit antithymocyte globulin (ATG, 88%), OKT3 (9%) or steroids (3%), and tacrolimus, mycophenolate mofetil and prednisone were used as maintenance immunosuppression. Diagnosis of AR AR was diagnosed by ultrasound-guided biopsy. Clinical indications for biopsy included persistent, unexplained increases in serum pancreatic enzymes (lipase and amylase) or plasma glucose. Surveillance protocol biopsies were obtained at 4 months, 1 year and 5 years after transplantation. For about half of the AR episodes (39 of 79), follow-up biopsies were performed about 1 month after treatment of AR episodes. Hematoxylin eosin-stained sections were reviewed by a specialist gastroenterology pathologist and graded using the University of Maryland (UMD) criteria initially and Banff criteria subsequently (17). All results are reported using Banff criteria. Antirejection treatment protocol In general, antirejection treatment was utilized for biopsy-confirmed Drachenberg grade 3 or greater rejection/banff grade 1 or greater ACR, whether diagnosed by clinically indicated or surveillance biopsy. Initial treatment consisted of corticosteroid boluses along with either OKT3 (5 mg/day for 7 10 days) or ATG (1.5 mg/kg/day for 5 10 days). Some cases showing Drachenberg Grade 2 rejection were also treated based on clinical judgment, using corticosteroid boluses with or without depleting antibody therapy. Corticosteroid boluses alone were used in patients who had contraindications for additional immunosuppressive therapy. For persistent AR, corticosteroid boluses alone, or additional courses of depleting antibody of variable duration, or no additional pulse therapy was used depending on clinical judgment. Chemoprophylaxis for CMV and Pneumocystis jiroveci infections was prescribed after initiation of treatment for rejection for 3 and 6 months, respectively (15). Definition of status of graft function Based on clinical and laboratory values, we grouped graft function into the following three categories: (1) full function (FF): C-peptide >200 pmol/l, fasting plasma glucose (FPG) and HbA1C normal or mildly elevated (FPG <126 mg/dl and HbA1c <7%), and not on any antihyperglycemic treatment; (2) partial loss (PL): C-peptide >200 pmol/l, and on oral antihyperglycemic agents or simple insulin therapy (one or two insulin injections per day); (3) complete loss (CL): CP <200 mg/dl, glycemic variability consistent with severe insulin deficiency, and on complex insulin therapy (>2 insulin injections per day or insulin pump). Statistical analyses Results are expressed as means ± standard deviation or median (interquartile range) for numeric variables, and counts and percentages for categorical variables. The incidence of AR episodes, incidences of PL or CL and patient survival were calculated by the Kaplan Meier (KM) method. Cox proportional hazards models were used to estimate hazard ratios and perform tests. Multivariable Cox model were used to assess the risk factors of AR and time-dependent Cox model was used to evaluate the impact of AR on graft function. Patients were censored at last follow-up in case they had not incurred graft failure. Multivariable models were derived using forward stepwise variable selection with a cut-off for inclusion of p = 0.05. History of rejection was entered into the Cox model as a time-dependent covariate to assess the associations between rejections and survivals. Statistical significance was assessed at the two-sided 0.05 level. All analysis was done by SAS 9.2 or higher software (SAS Institute, Cary, NC, USA). Results Demographic data Overall characteristics of the recipients and donors are reported in Table 1. Of the 227 PT, 56 (25%) were SPK, 102 (45%) PAK and 69 (30%) were PTA. Thus, 75% of the subjects had a solitary pancreas transplant. Mean age was 43 ± 9 years with 51% male, and median follow-up of 6.1 (IQR 3 9) years. CMV seromismatch status (donor positive/recipient negative) was present in 60 (26%) recipients. HLA mismatch >2/6 was found in 162 (73%) recipients and 10 of 193 recipients whose panel reactive antibody (PRA) was available had a PRA >20% at the time of PT. The 1020 American Journal of Transplantation 2013; 13: 1019 1025

Table 1: Demographic, baseline characteristics and length of follow-up of study patients AR and Pancreas Allograft Failure All SPK PAK PTA p Number of patients 227 56 102 69 Donor Age (years) 27 ± 12 29 ± 13 26 ± 12 28 ± 11 0.24 Males (%) 129 (57%) 34 (61%) 57 (56%) 38 (55%) 0.791 Recipients Age (years) 43 ± 9 45 ± 8 43 ± 9 41 ± 11 0.034 Males (%) 116 (51%) 32 (57%) 59 (58%) 25 (36%) 0.012 BMI (kg/m 2 ) 26.1 (5.2) 27.1 (5.8) 25.6 (4.4) 26 (5.6) 0.20 T1DM 212 (93%) 49 (88%) 96 (94%) 67 (97%) 0.92 Caucasian race 225 (99%) 56 (100%) 102 (100%) 67 (97%) 0.329 HLA mismatch >2/6 162 (73%) 39 (71%) 72 (74%) 51 (74%) 0.897 PRA% >20% 9 (5%) 2 (5%) 2 (2%) 5 (8%) 0.362 CMV status D+/R 60 (26%) 16 (29%) 28 (27%) 16 (23%) 0.756 Exocrine drainage 0.085 Primary bladder 140 (61%) 29 (52%) 62 (61%) 49 (71%) Primary enteric 87 (38%) 27 (48%) 40 (39%) 20 (29%) Induction treatment 0.058 ATG 200 (88%) 44 (79%) 95 (93%) 61 (88%) OKT3 21 (9%) 10 (18%) 4 (4%) 7 (10%) Others 6 (3%) 2 (3%) 3 (3%) 1 (1%) Follow-up (years) 6.1 (3 9) 6.1 (2 10) 6.0 (3 9) 6.3 (3 9) 0.86 Results are expressed as means ± standard deviation or median (interquartile range) for numeric variables, and counts and percentages for categorical variables. p-value of test of sameness of the three PT types based upon analysis of variance (ANOVA) or chi-square test. PTA group was younger than the SPK group (p = 0.014) and had more female recipients than both SPK and PAK groups (p = 0.02, and 0.004, respectively). Other baseline characteristics were similar among the three groups. Of the 227 PT, 78 (34%) had enteric conversion after a bladder drainage PT. The percentage of patients undergoing enteric conversion was higher in PTA group than that in SPK group (46% vs. 22%, p = 0.003). Incidence of treated pancreas AR Of the 227 recipients, 57 recipients developed 79 episodes of AR, and 19 recipients experienced more than one episode. The cumulative incidence for all AR episodes was 14.7% (10.7 19.8%), 19.7% (15.0 25.5%), 26.6% (20.9 33.3%) and 29.1% (22.9 36.1%) at 1, 2, 5 and 10 years. Figure 1 displays the AR incidence by PT categories. All the pancreas AR episodes occurred within 1 year in the SPK group (10%). One SPK recipient had an SPK rejection episode and the remainder had pancreas rejection alone. For PAK recipients, the cumulative incidence for pancreas AR was 9% (4.8 16.2%), 15.4% (9.6 24.0%), 26.6% (18.3 37.2%) and 30.2% (21.0 41.2%) at 1, 2, 5 and 10 years. The cumulative incidence for pancreas AR in PTA recipients was 25.0% (16.2 36.5%), 33.2% (23.0 45.3%), 39.2% (27.9 51.8%) and 42% (30.1-54.9%) at 1, 2, 5 and 10 years. Characteristics of all the episodes of treated pancreas AR Table 2 shows the characteristics of all treated episodes of AR. Of 79 episodes of AR, 23 (29%) were diagnosed American Journal of Transplantation 2013; 13: 1019 1025 Figure 1: Cumulative incidence of all the biopsy-confirmed treated pancreas AR episodes by PT type. by protocol biopsy. Per UMD 1997 classification, 61 (77%) were grade3ar,and7(9%)weregrade5ar.posttreatment biopsy was undertaken in 37 (47%) AR episodes, and 11 (14%) were treated with more than one course of therapy. Irreversible CL was seen in four recipients and PL in 10 recipients within 10 days of AR episodes in PTA and PAK recipients. Figure 2 shows the incidence of CL or at least PL within 10 days of AR. The cumulative incidence of at least PL within 10 days of AR was 0.45% (0.08 2.5%), 1021

Dong et al. Table 2: Summary of characteristics of all episode of AR All episodes SPK PAK PTA All episodes 79 6 32 41 Number of patients with more than one episodes 19 (33%) 0 (0) 6 (24%) 13 (50%) Reason for biopsy Protocol 23 (29%) 0 (0) 8 (25%) 15 (37%) Clinical 56 (71%) 6 (100%) 24 (75%) 26 (63%) AR grade DR2 8 (10%) 0 2 (6%) 6 (15%) ACR1 53 (67%) 6 (100%) 22 (69%) 25 (61%) ACR2 11 (14%) 0 6 (19%) 5 (12%) ACR3 7 (9%) 0 2 (6%) 5 (12%) Course of treatment 1 course 68 (86%) 5 (83%) 28 (88%) 35 (85%) > 1 course 11 (14%) 1 (17%) 4 (12%) 6 (15%) Posttreatment biopsy 37 (47%) 3 (50%) 17 (53%) 17 (41%) Figure 2: Cumulative incidence of at least PL, CL and at least PL or CL due to AR. 1.9% (0.8 4.9%), 4.9% (2.6 9.1%) and 8.2% (4.7 13.8%) at 1, 2, 5 and 10 years post PT. All four AR episodes resulting in CL were diagnosed more than 3 years post PT (median 3.4 years, range 3.0 5.6 years). One SPK recipient died with PL 2 days after the first AR episode (38 days post PT) due to sepsis and necrotizing fasciitis. Risk factors for pancreas AR In univariable analyses, as shown in Table 3, donor age (per 10 years) was associated with a higher risk of pancreas AR (hazards ratio [HR] = 1.34, 95% CI 1.09 1.65, p = 0.006), and recipient age (per 10 years) was associated with a borderline lower risk of AR (HR = 0.76, 95% CI 0.57 1.02, p = 0.065). PTA was associated with higher risk of pancreas AR (HR = 4.241, 95% CI 1.745 10.310, p = 0.001), while PAK transplant was associated with borderline higher risk (HR = 2.32, 95% CI 0.96 5.71, p = 0.061) and the risk was greater for the PTA group than the PAK group (p = 0.002). When these variables were en- Table 3: Baseline risk factors for AR obtained from univariable Cox regression HR (95% CI) Donor Age, per 10 years 1.341 (1.088, 1.6540) 0.006 Male gender 0.662 (0.394, 1.114) 0.120 Recipients Age, per 10 years 0.758 (0.565, 1.018) 0.065 Male gender 0.653 (0.386, 1.105) 0.112 CMV mismatch 1.052 (0.590, 1.875) 0.863 HLA mismatch (>2/6) 0.927 (0.525, 1.636) 0.793 PRA (>20%) 1.796 (0.649, 4.967) 0.259 Transplant type SPK (reference) 1 PAK 2.324 (0.961, 5.710) 0.061 PTA 4.241 (1.745, 10.310) 0.001 Induction therapy ATG 1.533 (0.211, 11.112) 0.673 OKT3 2.762 (0.345, 22.084) 0.338 Steroids (reference) 1 Exocrine drainage Primary bladder (reference) 1 Primary enteric 1.089 (0.570, 2.080) 0.795 Enteric conversion 0.844 (0.434, 1.643) 0.618 tered together in a Cox model, donor age and PTA and PAK transplant remained as significant risk associations with pancreas AR (HR = 1.40, p = 0.003; HR = 4.17, p = 0.002; HR = 2.46, p = 0.049, respectively). Univariable analysis showed that donor and recipient gender, CMV mismatch, HLA mismatch, PRA, induction drug, method of exocrine drainage and treated kidney AR were not associated with the risk of treated pancreas AR. Donor-specific antibodies (DSA) were tested in 30 recipients after 2007 and therefore could not be analyzed as a predictor for rejection in the full data set. Of these 30 recipients, five had treated AR; three recipients showed DSA for class II before AR, one had DSA for both class I and class II, and one had no antibodies. But when cut-off value of 1500 was used to identify DSA positive, none of those five patients was DSA positive. p 1022 American Journal of Transplantation 2013; 13: 1019 1025

AR and Pancreas Allograft Failure Impact of pancreas AR on long-term pancreas graft failure and patient survival The cumulative incidence of CL at 1, 2, 5 and 10 years was 3.1% (95% CI, 1.5 6.3%), 4.6% (2.5 8.3%), 14.0% (9.7 19.9%) and 20.4% (14.3 28.1%), and that of at least PL was 11.1% (7.6 15.9%), 16.0% (11.7 21.5%), 29.5% (23.4 36.3%) and 44.9% (36.3 53.8%) (Figure 2). Of the 57 recipients with AR, 23 had FF, 16 had PL and 18 had CL at last follow-up. In univariable analysis AR was associated with an increased risk for both CL (HR 3.86, 95% CI 1.87 7.99, p < 0.001) and at least PL (HR 2.32, 95% CI 1.37 3.94, p = 0.002). Multivariable Cox regression analysis was done to evaluate the impact of AR on long-term graft function by time period of first AR occurrence, and to account for the influence of potential confounders. The first AR episode beyond 3 months post PT was associated with an increased risk for CL (HR 3.79, 95% CI 1.75 8.22, p < 0.001). The first AR episode occurring during 3 24 months post PT was associated with a significantly increased risk for at least PL, with a higher HR for AR occurring during 12 24 months (HR 6.25, 95% CI 2.66 14.72, p < 0.001) than that for AR during 3 12 months (HR 2.84, 95% CI 1.24 6.49, p = 0.014). The first AR episode occurring within 3 or beyond 24 months after PT was not associated with at least PL (HR 0.97, 95% CI 0.29 3.23, p = 0.96; and HR 1.42, 95% CI 0.40 5.07, p = 0.59, respectively). In addition, multivariable analysis showed that recipient age (per 10 years) was also significantly associated with both CL and at least PL (HR = 0.91, p < 0.001; and HR = 0.97, p < 0.031, respectively). PRA >20% was related to an increased risk for CL (HR = 4.24, p = 0.020) and PAK transplant was significantly related to at least PL (HR = 2.95, p = 0.004). Donor age and gender, HLA mismatch, CMV mismatch, induction therapy, method of exocrine drainage and treated kidney rejection were not associated with the risk of pancreas graft failure. Patients with protocol biopsy identified rejection had an increased risk of complete graft failure (HR 4.62 95% CI 1.79 11.9) and a marginally significantly increased risk of partial failure (HR 2.17 95% CI 0.97 4.85) compared to those with clinical rejection. Of the entire cohort, 48 recipients died during followup. Overall patient survival at 1, 2, 5 and 10 years posttransplant was 95.6% (92.0 97.7%), 91.4% (86.9 94.4%), 82.8% (76.9 87.4%) and 71.3% (63.0 78.5%). Of 57 recipients with AR, 9 died during the study period. Treated pancreas AR was not a risk factor for mortality (HR = 0.87, 95% CI 0.42 1.82, p = 0.71). Impact of kidney transplant rejection on pancreas allograft function in the SPK and PAK cohorts The cumulative incidence of kidney transplant rejection in PAK recipients was 2.0% (0.5 6.9%), 3.1% (1.1 8.7%), 5.4% (2.3 12.1%) and 8.1% (3.5 17.6%) at 1, 2, 5 and 10 years. Among SPK recipients, it was 10.8% (5.0 21.6%) and 15.3% (7.9 27.6%) at 1 and 2 years, respectively, while no kidney transplant rejection was encountered after 2 years post PT. The incidence of kidney transplant American Journal of Transplantation 2013; 13: 1019 1025 rejection was borderline higher in SPK recipients than that in PAK recipients (p = 0.049, Log-rank test). However, the presence or absence of kidney transplant rejection did not result in higher pancreas allograft loss in both groups. Discussion This retrospective study focuses on the effect of treated pancreas AR on subsequent graft failure in three types of PT. Importantly, it is confined to the most recent era of PT with ATG induction for the majority and tacrolimus-based triple immunosuppression for all recipients. Furthermore, 23 (29%) surveillance biopsy-proven treated AR episodes were included in the analysis. Finally, we defined graft failure as complete or partial failure based on clinical manifestations and C-peptide concentrations. The main findings were that although the incidence of CL or at least PL due to treated pancreas AR was low, AR was associated with an increased risk of long-term CL and at least PL, and the time at which the first AR episode was observed influenced subsequent graft function. The impact of AR on long-term graft survival has only been reported for SPK recipients. Studies reported from Euro-SPK001 and Sweden (23-year single-center clinical experience) showed that AR was a significant prognostic factor for pancreas graft failure in SPK recipients (18,19). Our study confirmed these findings, and expanded them to all three PT categories. However, Reddy et al. recently reported that pancreas alone rejection did not affect the pancreas graft half-life in a 5-year follow-up after analyzing a large cohort of SPK reported to the United Network for Organ Sharing Database (UNOS) during 1988 1997 (20). Notably, the 1-year pancreas AR incidence in that study was only 3%, which was lower than that in our cohort (10% for SPK and 15% for all PT recipients) and another report using UNOS during 1995 2006 (14%) (21). One reason for this low incidence may be the absence of protocol pancreas allograft biopsies in many centers in that era, resulting in underestimation of the incidence of AR and AR-related graft loss. AR identified on surveillance protocol biopsies accounted for 29% of the AR episodes in the current study. In addition, graft survival in that study was estimated by graft half-life with the assumption of steady graft attrition after the first year and exponential distribution of survival time beyond 5 years. But visual inspection of the KM curve demonstrates that this may not always be the case and this will impact half-life calculations. A possible explanation for the negative effect of AR on long-term allograft function may be chronic rejection. A previous study reported that the most significant risk factor for chronic rejection was a previous AR episode (RR = 4.41, p < 0.001), and 91% of the recipients who lost their pancreas allograft to chronic rejection had been treated previously for AR (6). In our study, AR with chronic fibrosis or 1023

Dong et al. sclerosis was reported in two biopsies, of which one recipient developed immediate CL after AR, while other patient developed immediate PL followed by CL, 3 months later. Another important issue related to both AR and graft failure is donor-specific HLA antibody (DSA). De novo DSAs detected at AR were reported to be associated with reduced renal allograft survival (22 24). However, we were unable to reasonably assess the relationship between DSA, AR and pancreas allograft survival because of the small number of recipients with AR who had DSA testing. It has been reported in kidney transplantation that the time of AR occurrence could influence the impact of AR on longterm graft survival. The later the AR occurred, the higher risk of kidney graft failure (9). In a cohort of SPK recipients, pancreas survival was significantly higher among those without rejection than those with late rejection (first AR episode occurring >3 months posttransplantation) (25). In the present study, we found that a first AR episode occurring 3 24 months post PT was associated with increased risk for at least PL, with a higher risk associated with AR occurring after 12 months. Possible reasons for varied risks of graft function loss associated with time of AR are unclear. In kidney transplantation, it has been reported that recipient T cells recognizing donor class I molecules, presented by recipient antigen presenting cells, may be important mediators of late AR and chronic rejection (10). Late AR might signal ongoing immunological activity associated with a higher likelihood of chronic rejection and graft loss (6,26). In addition, late AR may be triggered by infections or other unknown factors, and nonadherence to immunosuppressive medication is more likely to affect late than early rejections (9). Since AR episodes beyond 12 months were observed only in PTA and PAK recipients, it is difficult to conclude that a first AR episode between 12 and 24 months is associated with higher risk of at least PL than that between 3 and 12 months for the individual PT types. Furthermore, the first AR episode occurring within 3 months as well as beyond 24 months was not associated with increased risk of at least PL. Further studies with larger number of all three PT categories are required to verify these findings and understand the possible underlying factors responsible for such associations. The incidence of AR was lower than previous studies, especially in PTA and PAK cohort, most likely related to our ATG induction and tacrolimus-based triple immunosuppression in the most recent era of PT (6). However, PTA was still a risk factor for AR. Another risk factor for AR in our study was donor age. Young donor pancreases were less prone to AR than those from older donors. Possible reasons for these findings could be decreased functional mass or preexisting atherosclerotic lesions of the older donor pancreas leading to impaired microcirculation and early ischemia reperfusion injury. This may further induce the release of inflammatory cytokines, activation of antigen presenting cells and immigration of recipient HLA class II-positive leukocytes into the transplants, subsequently triggering the AR (27). Recently, a Pancreas Donor Risk Index (PDRI) has been developed based on one transplant and 10 donor variables to assess the differential impact of organ quality on pancreas transplant outcomes. Among the donor factors, age appeared to be most important risk factor for graft failure. After adjusting for other factors, increase in donor age from 28 to 45 was associated with 56% increase in graft failure (28). This study has several limitations. First, it is a retrospective, observational cohort study. Second, it is a single-center study, limited by the number and geographic location of the study population. Third, the number of patients with biopsy-confirmed treated AR is relatively small among the subcategories of PT types. Lower incidence of AR in our cohort due to generally low immunologic risk population (99% Caucasian and only 5% with PRA > 20%), cannot be ruled out. Also, PRA data were available for limited number of patients. In summary, the current study demonstrates that although biopsy-confirmed treated AR as a cause of CL or PL was not common with current induction and maintenance immunosuppressive regimens in PT, it could have long-term negative effects on graft failure. Since about one third of AR episodes were diagnosed by protocol biopsies and associated with an increased risk of complete graft failure, we recommend serious consideration be given to performing protocol biopsies to improve graft survival, especially in PTA and PAK recipients. Our data need to be confirmed with prospective studies at other centers. Acknowledgments We thank patients and their families, contributing investigators and other study personnel who made the study possible. Ming Dong s research has been supported by promotive research fund for excellent young and middleaged scientists of Shandong Province (Grant number: BS2009YY014). This study was supported by NIH grant NCATS UL1 TR000135. Disclosure The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation. References 1. Gruessner AC, Sutherland DE. Pancreas transplant outcomes for United States (US) and non-us cases as reported to the United Network for Organ Sharing (UNOS) and the International Pancreas Transplant Registry (IPTR) as of June 2004. Clin Transplant 2005; 19: 433 455. 2. McCullough KP, Keith DS, Meyer KH, Stock PG, Brayman KL, Leichtman AB. Kidney and pancreas transplantation in the United States, 1998 2007: Access for patients with diabetes and endstage renal disease. Am J Transplant 2009; 9(4 Pt 2): 894 906. 1024 American Journal of Transplantation 2013; 13: 1019 1025

AR and Pancreas Allograft Failure 3. White SA, Shaw JA, Sutherland DE. Pancreas transplantation. Lancet 2009; 373: 1808 1817. 4. Leichtman AB, Cohen D, Keith D, et al. Kidney and pancreas transplantation in the United States, 1997 2006: The HRSA breakthrough collaboratives and the 58 DSA challenge. Am J Transplant 2008; 8(4 Pt 2): 946 957. 5. Drachenberg CB, Odorico J, Demetris AJ, et al. Banff schema for grading pancreas allograft rejection: Working proposal by a multidisciplinary international consensus panel. Am J Transplant 2008; 8: 1237 1249. 6. Humar A, Khwaja K, Ramcharan T, et al. Chronic rejection: The next major challenge for pancreas transplant recipients. Transplantation 2003; 76: 918 923. 7. Patil DT, Yerian LM. Pancreas transplant: Recent advances and spectrum of features in pancreas allograft pathology. Adv Anat Pathol 2010; 17: 202 208. 8. Joseph JT, Kingsmore DB, Junor BJ, et al. The impact of late acute rejection after cadaveric kidney transplantation. Clin Transplant 2001; 15: 221 227. 9. Opelz G, Dohler B. Influence of time of rejection on long-term graft survival in renal transplantation. Transplantation 2008; 85: 661 666. 10. Sijpkens YW, Doxiadis II, Mallat MJ, et al. Early versus late acute rejection episodes in renal transplantation. Transplantation 2003; 75: 204 208. 11. de Fijter JW. Rejection and function and chronic allograft dysfunction. Kidney Int 2010; 78(Suppl 119): S38 S41. 12. Stratta RJ. Late acute rejection episodes after vascularized pancreas transplantation. Transplant Proc 1998; 30: 1560 1561. 13. Stratta RJ. Late acute rejection after pancreas transplantation. Transplant Proc 1998; 30: 646. 14. Gloor JM, Winters JL, Cornell LD, et al. Baseline donor-specific antibody levels and outcomes in positive crossmatch kidney transplantation. Am J Transplant 2010; 10: 582 589. 15. Rostambeigi N, Kudva YC, John S, et al. Epidemiology of infections requiring hospitalization during long-term follow-up of pancreas transplantation. Transplantation 2010; 89: 1126 1133. 16. Issa N, Amer H, Dean PG, et al. Posttransplant lymphoproliferative disorder following pancreas transplantation. Am J Transplant 2009; 9: 1894 1902. 17. Drachenberg CB, Papadimitriou JC, Klassen DK, et al. Evaluation of pancreas transplant needle biopsy: Reproducibility and revision of histologic grading system. Transplantation 1997; 63: 1579 1586. 18. Malaise J, Arbogast H, Illner WD, et al. Simultaneous pancreaskidney transplantation: Analysis of rejection. Transplant Proc 2005; 37: 2856 2858. 19. Yamamoto S, Tufveson G, Wahlberg J, Berne C, Wadstrom J, Biglarnia AR. Factors influencing outcome of simultaneous kidney and pancreas transplantation: A 23-year single-center clinical experience. Transplant Proc 2010; 42: 4197 4201. 20. Reddy KS, Davies D, Ormond D, et al. Impact of acute rejection episodes on long-term graft survival following simultaneous kidney-pancreas transplantation. Am J Transplant 2003; 3: 439 444. 21. Kaplan B, West-Thielke P, Herren H, et al. Reported isolated pancreas rejection is associated with poor kidney outcomes in recipients of a simultaneous pancreas kidney transplant. Transplantation 2008; 86: 1229 1233. 22. Worthington JE, Martin S, Al-Husseini DM, Dyer PA, Johnson RW. Posttransplantation production of donor HLA-specific antibodies as a predictor of renal transplant outcome. Transplantation 2003; 75: 1034 1040. 23. Everly MJ, Everly JJ, Arend LJ, et al. Reducing de novo donorspecific antibody levels during acute rejection diminishes renal allograft loss. Am J Transplant 2009; 9: 1063 1071. 24. Terasaki PI, Ozawa M. Predictive value of HLA antibodies and serum creatinine in chronic rejection: Results of a 2-year prospective trial. Transplantation 2005; 80: 1194 1197. 25. Arbogast H, Malaise J, Illner WD, et al. Rejection after simultaneous pancreas-kidney transplantation. Nephrol Dial Transplant 2005; 20(Suppl 2): ii11 7, ii62. 26. Papadimitriou JC, Drachenberg CB, Klassen DK, et al. Histological grading of chronic pancreas allograft rejection/graft sclerosis. Am J Transplant 2003; 3: 599 605. 27. Tan M, Kandaswamy R, Sutherland DE, Gruessner RW, Gruessner AC, Humar A. Risk factors and impact of delayed graft function after pancreas transplants. Am J Transplant 2004; 4: 758 762. 28. Axelrod DA, Sung RS, Meyer KH, Wolfe RA, Kaufman DB. Systematic evaluation of pancreas allograft quality, outcomes and geographic variation in utilization. Am J Transplant 2010 2010; 10: 837 845. American Journal of Transplantation 2013; 13: 1019 1025 1025