The UK Pancreas Allocation Scheme for Whole Organ and Islet Transplantation

American Journal of Transplantation 2015; 15: 2443 2455 Wiley Periodicals Inc. C Copyright 2015 The American Society of Transplantation and the American Society of Transplant Surgeons doi: 10.1111/ajt.13284 The UK Pancreas Allocation Scheme for Whole Organ and Islet Transplantation A. Hudson 1,y, L. Bradbury 1,y, R. Johnson 1, S. V. Fuggle 1,2,J.A.M.Shaw 3, J. J. Casey 4, P. J. Friend 2 and C. J. E. Watson 5, * 1 Organ Donation and Transplantation Directorate, NHS Blood and Transplant, Bristol, England 2 The Oxford Transplant Centre, Churchill Hospital, Headington, Oxford, England 3 Institute of Cellular Medicine (Diabetes), The Medical School, Newcastle-upon-Tyne, England 4 Scottish Islet Transplant Unit, The Royal Infirmary, Edinburgh, Scotland 5 University Department of Surgery, Addenbrooke s Hospital, Cambridge and the NIHR Cambridge Biomedical Research Centre, England Corresponding author: Christopher Watson, cjew2@cam.ac.uk y Equal first authors. In order to develop a national allocation scheme for donor pancreases, factors affecting waiting time and transplant outcomes in the United States (US) and United Kingdom (UK) were analyzed and compared. Blood group, sensitization, dialysis requirement, and whether the patient was waiting for a kidney and pancreas or pancreas alone affected waiting time in both countries; ethnicity and body mass index (BMI) also affected waiting time in the US. Ninety-day pancreas survival was similar in the UK and US, and was poorer for patients receiving a pancreas alone, with older donors, higher BMI and longer duration of ischemia in both countries. Factors affecting outcome, together with published data on factors affecting islet transplantation, informed the development of a points based allocation scheme for deceased donor pancreases in the UK providing equitable access for both whole organ and islet recipients through a single waiting list. Analysis of the allocation scheme 3 years after its introduction in December 2010 showed that the results were broadly as simulated, with a significant reduction in the number of long waiting patients and an increase in the number of islet transplants. There remains a surplus of highly sensitized patients in the waiting list, which the scheme should address in time. Abbreviations: BMI, body mass index; crf, calculated reaction frequency: proportion of 10 000 blood group compatible UK donors against which the recipient has HLA-specific antibodies; DBD, donation after brain death; DCD, donation after cardiac death; HLA, human leucocyte antigen; NHS, National Health Service; NHSBT, National Health Service Blood and Transplant; ODT, Organ Donation and Transplantation Directorate of NHSBT; OPTN, Organ Procurement and Transplantation Network; PAK, pancreas after kidney transplant; PTA, pancreas transplant alone; SPK, simultaneous (combined) pancreas and kidney transplant; UK, United Kingdom; US, United States Received 05 October 2014, revised 01 February 2015 and accepted for publication 23 February 2015 Introduction Pancreas transplantation is an established treatment for patients with life-threatening complications of diabetes, but access to such transplants is variable. In the United Kingdom (UK), even though the first pancreas transplant was performed in 1979, the widespread adoption of pancreas transplantation had been slow, with only 32 pancreas transplants performed in 2000. In contrast, an estimate based on the proportion of young diabetic patients listed for kidney transplantation suggested 200 patients a year might benefit. In 2002, a task force was established to develop pancreas transplantation in the UK. This task force comprised representatives from the seven transplant centers performing pancreas transplants at the time. The group included representation from commissioners and UK Transplant (later renamed the Organ Donation and Transplantation [ODT] Directorate of National Health Service Blood and Transplant [NHSBT]). National Commissioning of Pancreas Transplantation was implemented in Scotland in 2000, and followed in England in 2004 and in Wales in 2006. The National Health Service (NHS) covers all the costs of the transplant and postoperative care, including immunosuppression, for the life of the patient. At that time, the task force devised a basic allocation scheme, which divided the UK into seven equal zones each centerd around a single transplant center; each center had the first choice of all deceased donor pancreases that were offered within their donation zone. Allocation of pancreases to patients on an individual center s waiting list was left to that center s discretion, with the stipulation that recipients should have an identical blood group to the donor. It was assumed that referrals for transplantation would follow a 2443

Hudson et al similar zonal pattern to donor organ allocation, but this did not happen with the result that, although each center had roughly a seventh of the national pancreas donors, some had much more than a seventh of the national waiting list; zones were therefore reviewed regularly and adjusted to accommodate these differences. The zones were further readjusted to accommodate addition of the Welsh pancreas transplant program. This zonal allocation scheme had several shortcomings. In spite of adjusting the zones, there remained a marked inequality of waiting times to transplantation between centers. The scheme resulted in each center having access to a limited pool of donors that was insufficient to address the increasing numbers of patients on the waiting list who were sensitized to human leucocyte antigens (HLA). Lastly there was no transparency and considerable variability in how pancreases were allocated within each center. To complicate allocation further, pancreatic islet transplantation had been nationally commissioned in England in 2008 and Scotland in 2009, but the three isolation laboratories were only accessing to pancreases that had been deemed unusable for whole organ transplantation, often with long ischemic times, and as a consequence they struggled to obtain satisfactory islet yields. The inequity of access to donor pancreases for both whole organ and islet transplantation prompted the development of a national patientspecific allocation scheme. A second task force was convened to examine available data and develop a scheme for the fair allocation of donor pancreases for both whole organ and islet transplantation; one that would afford potential recipients an equal opportunity of getting a pancreas and reduce variability between centers. As a prelude to this, the criteria for suitability for both islet and whole pancreas transplant waiting lists were reviewed. These criteria are regularly evaluated and updated, and are available on http://odt.nhs. uk/transplantation/guidance-policies/. Briefly, patients with insulin-dependent diabetes and renal failure (glomerular filtration rate less than 20 mls/min) are eligible for combined pancreas and kidney transplantation. Patients with life-threatening hypoglycemic unawareness are eligible for solitary pancreas or islet transplantation. Patients with a functioning kidney transplant and poor glucose control are also eligible for a pancreas or islet after kidney. Patients with type 2 diabetes must have a body mass index (BMI) less than 30 kg/m 2. An appeals panel was established to discuss listing of patients who fell outside of these criteria. The task force reviewed UK data on whole organ transplantation looking at factors associated with waiting time and outcomes following transplantation. Since the UK dataset was relatively small, data were also obtained from the Organ Procurement and Transplant Network (OPTN). The OPTN data were first compared to UK data to establish whether similar factors accounted for both the time waiting for a transplant and the outcomes following transplantation. Those factors identified in both United States (US) and UK datasets were then used to establish a national organsharing algorithm. There were insufficient data to use the same process for islet transplantation, so assumptions were made based on published data where available. This paper describes the analysis that underpinned the final UK pancreas allocation scheme, which was introduced in December 2010, and the results of that scheme over the first 3 years. Importantly, for the first time, organ allocation for both islet and whole organ recipients has been managed by means of a single waiting list. Materials and Methods The Pancreas Allocation Task Force The Pancreas Allocation Task Force comprised representatives of the whole organ transplant centers together with representatives of the islet purification laboratories, islet transplant centers histocompatibility and immunogenetics laboratories, and national commissioners together with statisticians from the Organ Donation and Transplantation Directorate of NHSBT. Pancreas transplant waiting list data for modeling OPTN data on 5101 adult patients joining the waiting list for a pancreas transplant in the US between 1 January 2005 and 31 December 2006 were compared with 398 adult patients joining the active pancreas transplant waiting list in the UK over the same time period. Survival analyses considered time from listing to receiving a transplant. Patients who died or were removed from the list within the first year, or were still waiting for a transplant at 1 year were censored at the relevant time point. The association between patient characteristics and estimated time to transplant of patients on the pancreas transplant list was investigated for both the UK and US data. Univariate analyses used the Kaplan Meier method of estimation and comparisons between groups of patients were made using the log-rank test. To examine fully the factors associated with waiting time before pancreas transplantation in both the UK and the US, independent multifactorial Cox proportional hazards regression models were developed. The hazard ratios from this analysis are interpreted as the relative chance of transplant within 1 year of listing. The baseline group for each factor has a relative chance of 1.0. For the purposes of the analysis, ethnicity was coded as white, nonwhite and not known, in order that the two national datasets could be compared. Pancreas transplantation data for modeling Data on 3568 first pancreas transplants donated by adults after brain death (DBD) (either transplanted alone or combined with a kidney) performed in the US between 1 January 2004 and 31 December 2008 were obtained from OPTN. These data were compared with the 667 first adult DBD donor pancreas transplants performed in the UK in the same time period. Survival analysis considered time from transplant to the earlier of pancreas graft failure or patient death. The transplant survival times of patients still alive with a functioning graft beyond 90 days were censored. To help 2444 American Journal of Transplantation 2015; 15: 2443 2455

The UK Pancreas Allocation Scheme maximize the number of transplants analyzed, follow-up was restricted to 90-day posttransplant outcome. The term solitary pancreas is used to denote either a pancreas transplanted alone (PTA) or pancreas transplanted after a previous kidney transplant (PAK). Islet transplant data Expert opinion within the UK was sought and the literature was reviewed, including annual reports of the Collaborative Islet Transplant Registry, to identify factors that were considered to influence outcomes following islet transplantation. Development of a national allocation scheme The results of the analyses informed the allocation task force about factors influencing waiting times and posttransplant outcomes and led to the agreement on those factors to be considered in an allocation scheme. Once the factors were agreed, different weightings were applied to reflect variable importance attached to the identified factors. Computer simulations of different allocation protocols and weightings of factors were then run using real data of historic donors and waitlisted patients to understand the benefits and limitations of a number of different algorithms. The proposed scheme was subject to a number of iterative changes until the final scheme was agreed. Evaluation of the national allocation scheme over the first 3 years Data on all patients on the UK national waiting list for a whole pancreas or islets for the 12 months before the scheme started were compared to those for the first 3 years of the scheme, up until 30 November 2013. Data for all transplants performed in the same time period were also compared. Graft failure for whole pancreas transplants was defined as a return to insulin or oral hypoglycemic agents; for islet transplants it was defined as a C-peptide level of less than 50 pmol/l 90 min after a standard mixed meal tolerance test. Reported graft failure for US patients had been used in analysis of the modeling cohort. Results Pancreas waiting list demographics in the modeling cohorts The demographics of the UK and US waiting lists were broadly similar (Table S1). In particular, there was no difference in patient sex, blood group, HLA sensitization and BMI. There was a significant difference in listing for a combined kidney and pancreas or a pancreas alone (including pancreas after kidney): 39% of patients were listed for a solitary pancreas in the US compared to 19% in the UK. Waiting time to pancreas transplantation in modeling cohorts The median waiting time for those receiving a pancreas transplant in the UK between 2004 and 2008 was 178 days (95% confidence interval [CI] 143 213) compared to 534 days (95% CI 501 567) in the US. Factors influencing time to transplant are shown separately for patients listed in the UK and US (Table 1). In both countries, patients of blood group A and AB were more likely to receive a transplant than patients of blood group O; in the UK blood group B was also associated with a shorter waiting time but not in the US. Sensitization to HLA was associated with a longer waiting time, patients being Table 1: Factors influencing chance of receiving a pancreas transplant within 1 year of listing in the UK and US, 2005 2006 Recipient risk factor UK data Relative chance of transplant (95% CI) US data Relative chance of transplant (95% CI) Recipient blood group O 1.00 1.00 A 1.82 (1.38 2.40) 1.41 (1.28 1.56) B 2.67 (1.69 4.18) 1.08 (0.92 1.26) AB 3.05 (1.74 5.34) 1.80 (1.44 2.26) Sensitization (%) (for every 10% increase in 0.85 (0.80 0.91) 0.84 (0.81 0.86) Dialysis requirement at registration calculated reaction frequency) None 1.00 1.00 Hemodialysis 1.30 (0.94 1.81) 0.84 (0.75 0.93) Peritoneal dialysis 1.54 (1.14 2.08) 0.85 (0.73 0.99) Graft type Pancreas only 1.00 1.00 Kidney/pancreas 1.65 (1.13 2.42) 1.48 (1.33 1.65) Ethnic origin White 1.00 1.00 Nonwhite 1.01 (0.63 1.61) 0.77 (0.68 0.87) BMI (for every 1.0 unit increase) 1.07 (0.99 1.15) 0.74 (0.67 0.83) Factors influencing a patient s chance of transplant were modeled separately for US and UK data using a Cox proportional Hazards regression model. p < 0.05. p < 0.01. p < 0.001. UK, United Kingdom; US, United States. American Journal of Transplantation 2015; 15: 2443 2455 2445

Hudson et al 15% less likely to receive a transplant for every 10% increase in calculated reaction frequency (crf, proportion of 10 000 blood group compatible UK donors against which the recipient has HLA-specific antibodies). Patients waiting for an isolated pancreas were also likely to wait longer in both countries than patients waiting for a combined kidney and pancreas transplant. In the US, patients on dialysis were less likely to receive a transplant than those not requiring dialysis; the opposite was true in the UK. In the US, patients were less likely to be transplanted as their BMI increased, and less likely also if they were nonwhite; there was no difference in the chance of being transplanted according to BMI or ethnicity in the UK, although most UK centers had an arbitrary BMI limit for listing of around 30 kg/m 2. Transplantation in the modeling cohorts Between 2004 and 2008 the annual number of pancreas transplants increased in the UK, while remaining relatively static in the US; there was no difference in 90-day transplant survival over that time (Table 2 with discussion in Supplementary Material). Univariate analyses Younger donor pancreases were associated with significantly better transplant survival both in the UK and the US compared to those from older donors (p < 0.05 UK; p < 0.001 US). There was also a significant effect of donor BMI on outcome in the UK with 96% transplant survival when the donor BMI was 19 and under, falling to 77% survival for donor BMI 30 34 (p < 0.01). No such trend was seen in US data. More pancreases were transplanted locally in the US than the UK (77% compared to 61%), and in both countries the 90-day pancreas transplant survival was better for pancreases retained locally, although the difference was only significant in the US (89% local compared to 87% for imported, p < 0.05). Solitary pancreas transplants (PTA and PAK) were more common in the US compared to the UK, but combined pancreas and kidney transplantation was the commonest form of pancreas transplant overall (87% in the UK, 71% in the US). The 90-day transplant survival following combined kidney and pancreas transplantation (SPK) was significantly better than following solitary pancreas transplantation in the UK (91% SPK, 85% PTA, 77% PAK, p < 0.01); the results of pancreas transplantation in UK recipients who had previously undergone a kidney transplant (PAK) were particularly poor. There was no difference in 90-day transplant survival in the US (89% SPK, 88% PTA, 87% PAK). As expected, ischemic time had a significant effect on 90-day transplant survival in both the UK and US. Where the ischemic time was under 9 h transplant survival was 96% in the UK, 91% in the US; where the cold ischemic time was beyond 16 h survival fell to 88% in the UK, 85% in the US (p < 0.01 UK; p < 0.05 US). In the US, more pancreases were transplanted within 9 h (27% US, 8% UK), possibly reflecting that more pancreases are transplanted locally in the US; more pancreases were also transplanted beyond 16 h in the US, (17% UK, 24% US) perhaps reflecting the greater distances involved when organs are shipped. Multivariate analysis of factors associated with 90-day pancreas transplant survival Separate Cox proportional hazards regression modeling of UK and US data identified four factors that significantly influenced outcome in both the UK and US data, namely whether the pancreas was transplanted alone or with the kidney, donor age, donor BMI, and ischemic time (Table 3). Factors associated with the outcome of islet transplantation A literature review and expert opinion identified younger donors with higher BMI, a short cold ischemic time, less than 6 months between first and second islet transplant, and good HLA matching to be desirable for optimal islet transplantation (1 6). The final allocation scheme Seven factors were included in the final allocation scheme implemented in the UK in December 2010. Each factor was given a different points-weighting and the person on the waiting list with the highest points total would be offered the transplant first. The factors were cold ischemic time, sensitization to HLA, dialysis status, waiting time, donorrecipient age matching, donor BMI, and donor-recipient HLA mismatch. Travel time by road from the donor hospital to the recipient s transplant center or islet isolation center was used as a surrogate for cold ischemic time such that the closest three transplant centers or closest islet isolation center had priority. For donation after cardiac death (DCD) donors, the travel time weighting was higher than DBD donors to minimize cold ischemia further. Pancreases were allocated to blood group identical recipients with the exception of recipients, who were blood group AB, who also had access to donors of blood group A (Table 4). Priority was given to highly sensitized patients (crf 75%) over nonsensitized patients, with an exponential weighting to increasingly raise their chances of a transplant with increasing sensitization. Increased weighting was also given for patients on dialysis, and for those who had waited longer. Islet recipients awaiting a second graft were prioritized further to try to ensure a second graft within a median of 6 months. Age was weighted to minimize the age difference between donor and recipient. Lastly, BMI was given a weighting such that donor pancreases with a lower BMI would be preferentially allocated to whole organ 2446 American Journal of Transplantation 2015; 15: 2443 2455

The UK Pancreas Allocation Scheme Table 2: Univariate Kaplan Meier analyses comparing 90-day pancreas transplant survival 1 January 2004 to 31 December 2008 UK data US data Factor N (%) % alive/functioning at 90 days (95% CI) N (%) % alive/functioning at 90 days (95% CI) Year of transplant 2004 67 (10) 90 (79 95) 880 (19) 89 (87 91) 2005 102 (15) 89 (81 94) 942 (20) 89 (87 91) 2006 141 (21) 92 (86 95) 943 (20) 87 (84 89) 2007 197 (30) 89 (84 93) 900 (20) 89 (87 91) 2008 160 (24) 88 (82 92) 937 (20) 89 (87 91) Donor sex Male 332 (50) 91 (88 94) 3154 (69) 89 (88 90) Female 335 (50) 87 (83 91) 1448 (31) 88 (86 89) Donor age 0 17 73 (11) 96 (88 99) 946 (21) 90 (88 92) 18 39 318 (48) 91 (87 94) 3064 (67) 90 (88 91) 40 49 192 (29) 85 (79 89) 519 (11) 81 (77 84) 50þ 84 (13) 88 (79 93) 73 (2) 81 (70 88) Donor BMI 1 0 19 75 (11) 96 (88 99) 612 (13) 89 (87 91) 20 24 353 (53) 91 (87 93) 2243 (49) 89 (88 91) 25 29 203 (30) 87 (82 91) 1382 (30) 87 (86 89) 30 34 30 (4) 77 (57 88) 301 (7) 87 (83 90) 35þ 6 (1) 67 (20 90) 64 (1) 88 (77 94) Donor CMV Negative 388 (57) 88 (85 91) 1758 (38) 87 (86 89) Positive 268 (40) 91 (87 94) 2844 (62) 89 (88 91) Not reported 11 (2) 91 (51 99) Shipping Local 407 (61) 90 (87 93) 3556 (77) 89 (88 90) Imported 260 (39) 88 (83 91) 1046 (23) 87 (84 88) Recipient sex Male 385 (58) 89 (86 92) 2793 (61) 89 (88 90) Female 282 (42) 89 (85 92) 1809 (39) 88 (86 89) Recipient age 18 39 287 (43) 89 (85 92) 1910 (42) 89 (88 91) 40 49 280 (42) 92 (88 95) 1773 (39) 88 (86 89) 50þ 100 (15) 83 (74 89) 919 (19) 89 (86 91) Dialysis status No 345 (52) 88 (84 91) 1369 (30) 88 (86 90) Yes 306 (46) 90 (86 93) 1396 (30) 89 (87 90) Not reported 16 (2) 100 ( ) 1837 (40) 89 (88 90) Sensitization 0to<30% 564 (83) 89 (87 92) 4306 (94) 89 (88 90) 30 to <75% 67 (10) 91 (81 96) 187 (4) 85 (78 89) 75 to 100% 36 (5) 86 (70 94) 109 (2) 84 (76 90) Transplant type SPK 577 (87) 91 (88 93) 3261 (71) 89 (88 90) Pancreas alone 20 (3) 85 (60 95) 467 (10) 88 (85 91) PAK 70 (10) 77 (65 85) 874 (19) 87 (85 89) Waiting time 1 (months) 0to<3 261 (39) 92 (88 94) 1223 (27) 90 (88 91) 3to<6 178 (27) 88 (83 92) 763 (17) 87 (85 89) 6to<12 117 (18) 89 (82 93) 976 (21) 89 (87 91) 12 or more 111 (17) 87 (79 92) 1640 (36) 88 (87 90) HLA A, B, and DR mismatches 5 6 242 (37) 89 (84 92) 2605 (57) 89 (88 90) 0 4 425 (63) 90 (86 92) 1997 (43) 88 (87 89) (Continued) American Journal of Transplantation 2015; 15: 2443 2455 2447

Hudson et al Table 2: Continued UK data US data Factor N (%) % alive/functioning at 90 days (95% CI) N (%) % alive/functioning at 90 days (95% CI) Ischemia time 1,2 (hours) 0to<9 56 (8) 96 (87 99) 1238 (27) 91 (90 93) 9to<12 159 (24) 94 (89 97) 1098 (24) 89 (87 91) 12 to <16 340 (51) 87 (83 90) 1155 (25) 88 (86 90) 16þ 112 (17) 88 (80 92) 1111 (24) 85 (83 87) 1 Missing observations were imputed with the median. 2 For UK pancreas transplants cold ischemic time (cold perfusion in donor to reperfusion in recipient) was recorded, while for US transplants the total duration of ischemia was reported. p < 0.05. p < 0.01. p < 0.001. BMI, body mass index; CMV, cytomegalovirus; DR, D related; HLA, human leucocyte antigen; PAK, pancreas after kidney transplant; SPK, simultaneous (combined) pancreas and kidney transplant. transplantation and those with a higher BMI would be allocated to islet transplantation, since low BMI donors tend to give poorer islet yields (3,7). The effect of the new allocation scheme on the waiting list Table 5 details the demographics of the waiting list at the start of the new allocation scheme, and its predicted and actual composition after 3 years. Table 6 shows the median waiting times according to demographic factors at the two time points. The simulations had suggested that discrepancies in waiting times should be resolved by the end of 3 years. There was a marked reduction in the number of long waiting patients, with only 31% waiting more than a year compared to 54% at the introduction of the scheme. While an improvement, the effect on long waiting patients was not as great as the 24% predicted by the simulation. Part of this is accounted for by a relatively poor effect on the highly sensitized patients whose median wait fell from 714 to 441 days, while the nonsensitized and lesser-sensitized patients had waiting times nearer to the median. The proportion of highly sensitized patients fell from 36% to 34%, far short of the predicted 17% (Table 5). An increase in the proportion of older patients listed in the 3 years of the scheme meant there was a greater proportion of older patients than expected from previous practice, but the waiting times of the different age groups was similar. The proportions of patients of each blood group on the waiting list did not follow expectations, with an increase in blood group O patients (51 59%) and a reduction in group A patients (40 36%) and group B patients (9 4%). The discrepancies in blood group distribution were not explicable by differences in the new registrants, where blood group O was proportionately less (41%), but rather by the distribution of blood groups in the donors. The differences in the waiting times of patients at each of the whole organ centers also reduced (Figure 1), as did the median wait for a first islet transplant, which was now similar to the waiting time for a whole organ (median 203 days for islet, compared to 221 for pancreas). The median time on the list for those awaiting a second islet transplant was 170 days, close to the target of 180 days, but Table 3: Factors influencing 90-day pancreas transplant survival in the US and UK, 2004 to 2008 Risk factor UK data Relative risk (95% CI) of failure/death US data Relative risk (95% CI) of failure/death Transplant type SPK 1.00 1.00 Pancreas only 2.23 (1.36 3.99) 1.19 (1.00 1.44) Donor age 10-year increase 1.22 (1.00 1.49) 1.30 (1.20 1.41) Donor BMI 1 kg/m 2 increase 1.08 (1.02 1.14) 1.00 (0.98 1.02) Ischemia(1) 1 h increase 1.06 (1.00 1.13) 1.03 (1.02 1.04) Waiting time 1 month increase 1.02 y (1.00 1.04) 1.00 (0.99 1.01) Statistical significance: y p < 0.1; p < 0.05; p < 0.01; p < 0.001. 1 For UK pancreas transplants cold ischemic time was recorded while for US transplants the total duration of ischemia was reported. BMI, body mass index; SPK, simultaneous (combined) pancreas and kidney transplant; UK, United Kingdom; US, United States. 2448 American Journal of Transplantation 2015; 15: 2443 2455

Table 4: Factors and their weightings used for the allocation of pancreases in the 2010 UK allocation scheme The UK Pancreas Allocation Scheme Factor Contribution to final score (points) Travel time (surrogate for ischemic time) DBD donors Closest three transplant centers 365 Center outside the closest three 0 Islet patient (by default) 1 365 Priority islet patient 1 0 DCD donors Closest center 10 000 Next closest two centers 5000 Center outside the closest three 0 Within 150 miles of isolation laboratory 1 10 000 Sensitization Calculated reaction frequency) (%crf) 3 /1000 Dialysis On dialysis 180 Not on dialysis 0 Waiting time Whole organ and first islet graft Waiting ðdaysþ2 0 þ Patients listed for a second islet graft 0 þ 365 Waiting time ðdaysþ2 44:4 Donor recipient age matching Age difference ðyearsþ 2 Donor BMI (rounded to nearest whole number) Whole organ Islet 22 or less þ730 730 23 to 25 þ365 365 26 to 28 0 0 29 to 31 365 þ365 32 or over 730 þ730 Total HLA-A, B, and DR mismatch 0 to 4 mismatches 750 5 and 6 mismatches 0 HLA points adjustment for patients listed for first islet graft 2 0 mismatches 0 1 to 2 mismatches 150 3 to 4 mismatches 350 5 to 6 mismatches 700 Donor pancreases were to be transplanted into blood group identical recipients, except that a blood group AB recipient could receive a blood group A pancreas. This was revised in 2014 to allow highly sensitized (crf > 75%) patients of blood group B to receive blood group O kidneys. 1 Since islets are isolated from the pancreas at one of three national laboratories before culture and shipping to a recipient site for transplantation, traveling time was to the isolation laboratory, not the recipient center. Patients listed for a routine islet transplant gain points only if an isolation laboratory is within 150 miles of the donor hospital. Priority patients already receive extra points and do not require supplementary points. 2 This was designed to minimize sensitization following the first graft and so lessen the difficulty in getting the second graft which most recipients were anticipated to require. BMI, body mass index; DBD, donation after brain death; DCD, donation after cardiac death; UK, United Kingdom. 13:9 with an interquartile range that extended to 647 days, although the number awaiting a retransplant was small. The effect of the new allocation scheme on pancreas and islet transplantation Transplant activity increased in each year after the introduction of the scheme, from 207 in the year preceding to 252 in the third year. Islet transplantation grew from 5% to 11% of all pancreas transplants. As expected, the first 2 years of the scheme saw large numbers of long waiting American Journal of Transplantation 2015; 15: 2443 2455 patients transplanted, with 43% of all transplants in the first year going to patients waiting for more than 2 years; there were still proportionately more long waiters transplanted in the third year of the scheme, but an excess still remained on the waiting list at the end of the third year (Table 5, 9% waiting more than 2 years against a predicted 5%). High transplant rates for long waiting patients may explain why pancreases were still being transplanted at centers remote from the donor hospital with less well-matched grafts, reflecting the relatively heavy weighting for waiting time relative to travelling time and HLA-mismatch. The 2449

Hudson et al Table 5: Demographic details of patients on the national waiting list for whole organ and islet transplants before the introduction of the 2010 allocation scheme together with the predicted and actual composition 3 years later Waiting list prior to introduction of the scheme (1 December 2010) Actual waiting list 3 years after implementation of the scheme (30 November 2013) Simulated waiting list prediction at 3 years New registrations on the waiting list between 1 December 2010 and 31 November 2013 Number of patients 326 277 255 868 Factor N (%) N (%) (%) N (%) Time on waiting list (months) 0to<3 56 (17) 60 (24) (23) 3to<6 35 (11) 51 (20) (21) 6to<12 59 (18) 67 (26) (31) 12 to <24 108 (33) 55 (22) (19) 24 or more 68 (21) 22 (9) (5) Sensitization (crf%) 0to<10 210 (64) 168 (66) (83) 693 (80) 10 to <30 26 (8) 15 (6) (4) 36 (4) 30 to <75 50 (15) 26 (10) (4) 58 (7) 75 or more 40 (12) 46 (18) (9) 81 (9) On dialysis No 171 (52) 147 (58) (55) 567 (65) Yes 152 (47) 107 (42) (45) 301 (35) Not known 3 (1) 1 (0) (0) 0 (0) Transplant type SPK 252 (77) 195 (76) (73) 595 (69) PAK, PTA 49 (15) 36 (14) (18) 154 (18) Routine islet 21 (6) 18 (7) (7) 73 (8) Priority islet 4 (1) 6 (2) (2) 46 (5) Sex Male 177 (54) 120 (47) (51) 460 (53) Female 149 (46) 135 (53) (49) 408 (47) Blood group O 165 (51) 150 (59) (52) 400 (46) A 132 (40) 92 (36) (36) 347 (40) B 28 (9) 10 (4) (8) 86 (10) AB 1 (0) 3 (1) (2) 35 (4) Recipient age (years) Less than 30 12 (4) 16 (6) (9) 65 (7) 30 to <40 99 (30) 71 (28) (42) 264 (30) 40 to <50 132 (40) 101 (40) (34) 324 (37) 50 to <60 71 (22) 60 (24) (14) 188 (22) 60 or older 12 (4) 7 (3) (1) 27 (3) Ethnicity White 303 (93) 227 (89) (93) 774 (89) Nonwhite 23 (7) 23 (9) (7) 85 (10) Not known 0 (0) 5 (2) (0) 9 (1) crf, calculated reaction frequency: proportion of 10 000 blood group compatible UK donors against which the recipient has HLA-specific antibodies; PAK, pancreas after kidney transplant; PTA, pancreas transplant alone; SPK, simultaneous (combined) pancreas and kidney transplant. conversion rate from pancreas offers to actual transplants increased for DBD transplants (from 33% to 37%), while the rate of conversion for DCD donor transplants fell, probably because of an increase in offers of older DCD donors (the median age of UK DCD donors is now 55 years) (Table 7). Pancreas transplant outcomes under the new scheme The outcomes of pancreas transplants under the new scheme confirmed the expectation of inferior results for the first 2 years as a backlog of long waiting patients was transplanted, with subsequent improvement (Figure 2). 2450 American Journal of Transplantation 2015; 15: 2443 2455

The UK Pancreas Allocation Scheme Table 6: Median waiting time on the list before and 3 years after introduction of the national pancreas allocation scheme according to demographic factors 1 December 2010 1 December 2013 Factor N Median IQ range N Median IQ range All patients 326 402 166 668 255 221 97 402 Sensitization (crf%) 0to<10 210 358 150 571 168 201 75 368 10 to <30 26 455 353 628 15 167 62 278 30 to <75 50 467 158 739 26 185 116 344 75 or more 40 714 251 1130 46 441 151 1002 On dialysis No 171 368 151 650 147 173 75 359 Yes 152 445 179 676 107 302 116 472 Transplant type SPK 252 427 184 698 195 221 102 402 PAK, PTA 49 349 62 694 36 240 104 540 Routine islet 21 267 170 427 18 203 73 359 Priority islet 4 283 160 513 6 170 140 647 Sex Male 177 413 173 661 120 186 71 385 Female 149 397 161 694 135 269 116 419 Blood group O 165 400 166 713 150 278 117 425 A 132 409 171 622 92 174 70 378 B 28 399 112 613 10 54 19 122 AB 1 46 46 46 3 302 272 318 Recipient age (years) Less than 30 12 293 49 654 16 228 120 371 30 to <40 99 404 158 613 71 187 68 358 40 to <50 132 374 153 690 101 272 111 411 50 to <60 71 433 230 714 60 190 102 406 60 or older 12 433 244 692 7 405 313 647 Ethnicity White 303 400 161 683 227 206 95 405 Nonwhite 23 409 176 661 23 277 87 388 Whole Pancreas Transplant Center Cardiff 8 114 51 541 14 169 54 262 Edinburgh 32 401 170 524 25 320 221 506 Cambridge 9 130 28 410 20 245 102 366 Guy s 19 222 124 621 32 202 68 405 Manchester 61 466 278 795 47 186 103 425 Newcastle 10 322 145 512 11 116 53 312 Oxford 136 430 191 712 68 272 140 410 WLRTC 26 355 118 714 14 143 45 368 Islet Transplant Center Edinburgh 3 170 51 191 9 151 94 354 Royal Free 3 153 100 1395 2 87 2 171 King s College 1 294 294 294 3 318 36 360 Manchester 2 331 272 389 0 Newcastle 7 378 267 521 8 186 114 307 Oxford 9 205 166 489 2 746 647 844 crf, calculated reaction frequency: proportion of 10 000 blood group compatible UK donors against which the recipient has HLA-specific antibodies; PAK, pancreas after kidney transplant; PTA, pancreas transplant alone; SPK, simultaneous (combined) pancreas and kidney transplant. Discussion This paper illustrates how the 2010 UK pancreas allocation scheme was developed using UK and US data on waiting list and transplant outcomes to inform the development of a national algorithm. The analysis showed that waiting list American Journal of Transplantation 2015; 15: 2443 2455 demographics and transplant outcomes were broadly similar in the US and UK. In the waiting time analysis, transplant type, blood group, HLA sensitization, and dialysis predicted waiting time in both countries. However, some differences in practice 2451

Hudson et al final scheme with the ultimate aim being that for any given donor an ordered list of all suitable recipients on the national waiting list could be produced, the organ being offered to the highest ranking patients in succession until accepted. Patients awaiting an islet transplant were on the same national waiting list. The goal was to allow equal access to a transplant while minimizing differences in waiting time and optimizing transplant outcomes. Figure 1: Center waiting times. Scatter plot showing variation in center waiting times for whole organ before and each year after the introduction of the national allocation scheme. were apparent, such as the greater proportion of patients listed for a solitary pancreas transplant in the US compared to the UK. This may reflect the shorter waiting time for a combined kidney and pancreas transplant in the UK, since pancreas patients have priority for a kidney above all patients on the kidney waiting list other than certain 0-0-0 HLA-mismatched patients: children, HLA-DR homozygous adults and highly sensitized adults. Without this priority for a kidney, there may have been a preference for an early kidney alone transplant, particularly a live donor transplant, followed later by pancreas transplant. This would avoid enduring a longer wait on dialysis for a combined kidney and pancreas transplant with its adverse consequences (8,9). Having preferential access to a kidney for a combined kidney and pancreas transplant is not without consequences; it may result in potential living donor transplants being lost and will result in longer waiting times for patients awaiting a kidney alone. This is the cause of much debate within the UK and beyond. It is justifiable, in part, by the observation that survival from transplant is the same for kidney and SPK recipients, while survival from listing is poorer for those patients awaiting an SPK (data not shown). Factors affecting the outcomes of transplantation were also modeled, and ischemic time, donor age, donor BMI, and recipient waiting time were found to be significant both in UK and US datasets. Donor age, BMI, and cold ischemic time were also found to be significant factors in a subsequent analysis by Axelrod et al in deriving a Pancreas Donor Risk Index using data from the Scientific Registry of Transplant Recipients (10). They also found gender, race, height, cause of death, serum creatinine, and DCD donation to be risk factors; two of these, gender and race (ethnicity) were considered in our study but not found to be significant. The main stimulus for performing this comparison of UK and US pancreas transplantation was to provide data to inform a future UK pancreas allocation scheme. The final allocation scheme aimed to deliver equity of access to transplantation while optimizing graft outcome where possible. Each of the factors identified were reviewed by an expert panel and given an appropriate weighting in the The elements of the scheme have been described above. One component for which there was little evidence was a factor for HLA matching. Nevertheless better matching was likely to limit sensitization to HLA (2), something that might be important for a whole organ recipient needing a second transplant and is also important for someone awaiting an islet transplant (1), where multiple grafts may be required. The results of the first 3 years of the national allocation scheme are also presented here. The effect on the national waiting list was broadly as expected from the computer simulations. The magnitude of the effect on long-waiting patients was not as great as predicted, and may reflect the fact that, although long-waiting patients may be offered donor pancreases, the organs may either not be accepted for that patient or be deemed unsuitable on inspection at the recipient center. Hence, it appears that the scheme will take longer than the anticipated 3 years to achieve its aim. Having an excess of long waiting patients still on the waiting list has consequences on allocation of organs since waiting time is a dominant factor in the allocation scheme. A patient who has a crf of 75% is given a weighting equivalent to 392 days on the waiting list, for example, so while there remains an excess of long-waiting patients they may still get priority over sensitized patients, depending on waiting times and degrees of sensitization. The anticipated benefits of improved HLA mismatching and allocation of pancreases to centers near the donor hospital have not been fully realized, almost certainly for the same reasons. The scheme will be reviewed again with a view to increasing priority for sensitized patients if the excess persists. There is still a variation in waiting times between centers, but data suggest in part that this reflects different appetites for using older donors and DCD donors. The incorporation of the national islet transplant program into the allocation scheme has resulted in improved access to good quality pancreases and an increase in transplant activity; notably it did not result in a fall in whole organ transplantation, which rose from 196 transplants before the scheme to 223 three years later. One concern was that, although priority for a second transplant within 180 days was largely achieved, there remained some patients who had to wait much longer for their second graft. The scheme has subsequently been modified to aim for a shorter median waiting time of 3 months to a second islet graft. 2452 American Journal of Transplantation 2015; 15: 2443 2455

The UK Pancreas Allocation Scheme Table 7: Demographics of patients transplanted in the year before and the 3 years following the introduction of the 2010 National Pancreas Allocation Scheme Transplants Listing Year prior to scheme introduction New pancreas allocation scheme Year 1 results Year 2 results Year 3 results Predicted (end of Year 3) Patients listed in Years 1 3 Number of transplants 207 218 241 252 263 868 Waiting time (months) 0to<3 47 (23) 28 (13) 37 (15) 32 (13) (6) 3to<6 26 (13) 20 (9) 26 (11) 28 (11) (10) 6to<12 29 (14) 22 (10) 40 (17) 59 (23) (30) 12 to <24 63 (30) 55 (25) 94 (39) 106 (42) (52) 24 or more 42 (20) 93 (43) 44 (18) 27 (11) (1) Sensitization (crf%) 0to<10 157 (76) 155 (71) 168 (70) 180 (71) (83) 693 (80) 10 to <30 13 (6) 19 (9) 20 (8) 22 (9) (3) 36 (4) 30 to <75 16 (8) 30 (14) 28 (12) 29 (12) (7) 58 (7) 75 or more 21 (10) 14 (6) 25 (10) 21 (8) (8) 81 (9) Total HLA mismatch (A, B, DR) 0 to 4 136 (66) 162 (74) 203 (84) 224 (89) (94) 5 or 6 71 (34) 56 (26) 38 (16) 28 (11) (6) Center location Closest three 158 (76) 137 (63) 150 (62) 168 (67) (87) Other 49 (24) 81 (37) 91 (38) 84 (33) (13) On dialysis No 118 (57) 120 (55) 135 (56) 133 (53) (47) 567 (65) Yes 86 (42) 97 (44) 105 (44) 119 (47) (53) 301 (35) Not known 3 (1) 1 (0) 1 (0) 0 (0) 0 (0) Transplant type SPK 159 (77) 156 (72) 168 (70) 194 (77) (76) 595 (69) PAK, PTA 37 (18) 38 (17) 36 (15) 29 (12) (16) 154 (18) Routine islet 6 (3) 15 (7) 25 (10) 16 (6) (5) 73 (8) Priority islet 5 (2) 9 (4) 12 (5) 13 (5) (3) 46 (5) Sex Male 115 (56) 128 (59) 122 (51) 144 (57) (54) 460 (53) Female 92 (44) 90 (41) 119 (49) 108 (43) (46) 408 (47) Blood group O 97 (47) 83 (38) 103 (43) 113 (45) (49) 400 (46) A 84 (41) 104 (48) 105 (44) 98 (39) (45) 347 (40) B 21 (10) 24 (11) 24 (10) 31 (12) (3) 86 (10) AB 5 (2) 7 (3) 9 (4) 10 (4) (2) 35 (4) Recipient age (years) <30 12 (6) 10 (5) 11 (5) 19 (8) (11) 65 (7) 30 to <40 63 (30) 64 (29) 66 (27) 79 (31) (38) 264 (30) 40 to <50 91 (44) 83 (38) 98 (41) 89 (35) (35) 324 (37) 50 to <60 34 (16) 54 (25) 57 (24) 58 (23) (16) 188 (22) 60 7 (3) 7 (3) 9 (4) 7 (3) (1) 27 (3) Ethnicity White 196 (95) 201 (92) 218 (90) 224 (89) (92) 774 (89) Nonwhite 11 (5) 17 (8) 22 (9) 27 (11) (8) 85 (10) Not known 1 (0) 1 (0) 9 (1) Figures in brackets are percentages of groups. crf, calculated reaction frequency: proportion of 10 000 blood group compatible UK donors against which the recipient has HLA-specific antibodies; HLA, human leucocyte antigen; PAK, pancreas after kidney transplant; PTA, pancreas transplant alone; SPK, simultaneous (combined) pancreas and kidney transplant. In summary, analysis of factors affecting waiting times and transplant outcomes in the UK and US has allowed development of a patient-based, whole organ and islet transplantation national allocation scheme for donor pancreases in the UK. The actual effects of the scheme on waiting times for transplantation overall, and for subgroups such as sensitized patients, are broadly as American Journal of Transplantation 2015; 15: 2443 2455 predicted but it has taken 3 years to address most of the inequities present before the scheme was introduced. The long-term effect of the new allocation scheme on outcomes of transplantation remain to be seen, with early results predictably less good than previously as a greater number of long waiting patients have been transplanted. 2453

Hudson et al Figure 2: Graft and patient survival before and after introduction of the national allocation scheme. Author Contributions Alex Hudson and Lisa Bradbury performed the statistical analysis of the US and UK datasets. Together with Rachel Johnson they developed the allocation model and produced simulated allocations. Susan Fuggle, James Shaw, John Casey, and Peter Friend reviewed the data, advised on the simulations, and reviewed the final manuscript. Christopher Watson chaired the allocation working party and wrote the manuscript. Acknowledgments This work was supported in part by Health Resources and Services Administration contract 231-00-0115. The content is the responsibility of the authors alone and does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the US Government. We would like to acknowledge the contribution of Professor Chris Rudge for establishing the Pancreas Advisory Group and its predecessor, the Pancreas Task Force, which were instrumental in developing pancreas transplantation in the UK. We would also like to acknowledge the work of the working party involved in the development of the UK allocation scheme, which comprised Chris Watson (Chair), Peter Friend, Susan Fuggle, Alex Gimson, Edmund Jessop, Alex Hudson, Rachel Johnson, Susan Martin, Stephen Powis, James Shaw, and John Taylor. Disclosure The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation. 2454 American Journal of Transplantation 2015; 15: 2443 2455

The UK Pancreas Allocation Scheme References 1. Campbell PM, Salam A, Ryan EA, et al. Pretransplant HLA antibodies are associated with reduced graft survival after clinical islet transplantation. Am J Transplant 2007; 7: 1242 1248. 2. Campbell PM, Senior PA, Salam A, et al. High risk of sensitization after failed islet transplantation. Am J Transplant 2007; 7: 2311 2317. 3. Matsumoto I, Sawada T, Nakano M, et al. Improvement in islet yield from obese donors for human islet transplants. Transplantation 2004; 78: 880 885. 4. Nano R, Clissi B, Melzi R, et al. Islet isolation for allotransplantation: Variables associated with successful islet yield and graft function. Diabetologia 2005; 48: 906 912. 5. Collaborative Islet Transplant Registry Research Group. 2007 update on allogeneic islet transplantation from the Collaborative Islet Transplant Registry (CITR). Cell Transplant 2009; 18: 753 767. 6. Ponte GM, Pileggi A, Messinger S, et al. Toward maximizing the success rates of human islet isolation: Influence of donor and isolation factors. Cell Transplant 2007; 16: 595 607. 7. Briones RM, Miranda JM, Mellado-Gil JM, et al. Differential analysis of donor characteristics for pancreas and islet transplantation. Transplant Proc 2006; 38: 2579 2581. 8. Ojo AO, Meier-Kriesche HU, Hanson JA, et al. The impact of simultaneous pancreas kidney transplantation on long-term patient survival. Transplantation 2001; 71: 82 90. 9. Reddy KS, Stablein D, Taranto S, et al. Long-term survival following simultaneous kidney pancreas transplantation versus kidney transplantation alone in patients with type 1 diabetes mellitus and renal failure. Am J Kidney Dis 2003; 41: 464 470. 10. 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; 10: 837 845. Supporting Information Additional Supporting Information may be found in the online version of this article. Supplementary Materials Table S1: Demographics of patients on the pancreas transplant waiting list in the US and UK, 1 January 2005 31 December 2006. Table S2: Use of pancreases from solid organ donors. American Journal of Transplantation 2015; 15: 2443 2455 2455