Donation after brain death (DBD) is the predominant source of organs for transplantation, even though brain death accounts for only a small percentage of all-cause mortality in the United States.1 In contrast, cardiovascular death is a leading cause of mortality.2 Therefore, donation after cardiac death (DCD) represents an attractive strategy for remedying organ shortage and improving transplant wait-list mortality.3 The transplantation of DCD organs has risen dramatically over the last decade in response to federal mandates issued by the Health Resources and Services Administration and the Centers for Medicare and Medicaid Services, and 10.6% of all organ transplants currently use DCD organs.1
The utilization of livers from DCD donors for transplantation declined by nearly 10% from 2007 to 2008, and this decline could undermine the goal set by the Organ Donation and Transplantation Collaborative: 10% of all donors from DCD sources.1 This downward trend in DCD liver utilization likely stems from recent reports demonstrating inferior outcomes with DCD liver transplants versus DBD liver transplants.4-10 In particular, DCD livers are marred by higher rates of graft failure,11 retransplantation,8 and patient mortality,12 which are due at least in part to biliary complications (primarily ischemic cholangiopathy).13
In this issue of Liver Transplantation, Taner et al.14 report graft and patient survival data for DCD liver transplants similar to DBD liver transplants at the Mayo Clinic in Jacksonville. Although the authors should be commended for achieving such favorable outcomes, other single-institution experiences have also failed to identify differences in graft5, 15-17 and patient survival rates3, 4, 16-18 between DCD and DBD livers because of limited sample sizes. The study by Taner et al. was powered to resolve a 10% or greater differential in graft survival, which is admittedly greater than the differences observed between DCD and DBD livers in many studies.5, 15, 16 Consequently, although statistical significance was not achieved, there were clearly disparate trends in graft survival for DCD and DBD grafts. A peculiar feature of this study is the fact that although DCD livers yielded an inferior graft survival rate, they were associated with a superior patient survival rate in comparison with DBD livers.14 Because this result could have been achieved only with aggressive retransplantation, the local availability of livers for retransplantation in Jacksonville may help to explain this outcome. Therefore, it is not clear whether these results are generalizable to other localities.
Despite the debate about the relative merits of census (registry) data versus sample (single-center) data, the national registry reliably captures both retransplantation and patient mortality data because of the linkage with the Social Security Death Master File. As for issues of sample size, analyses of the national registry have clearly demonstrated worse graft survival3, 7, 8, 11, 19, 20 and, more recently, worse patient survival.12 In a complementary manner, single-institution reports provide important insights into granular covariates (eg, asystole/cross-clamp times)14 and outcomes (eg, ischemic cholangiopathy)4 not reported to the registry, but they are often subject to limitations in sample size. However, inadequate power can be addressed through the application of specific statistical methodologies. For instance, using a meta-analysis of sample data provided by 11 single-institution observational studies representing 489 DCD recipients and 4455 DBD recipients, we also identified inferior graft and patient survival rates for DCD recipients.13
Another plausible explanation for the comparable results for DCD and DBD livers observed by Taner et al.14 is that they employed stringent acceptance criteria for DCD liver transplantation. Accordingly, the DCD donors were younger (40 versus 47 years), were less likely shares (44% versus 78%), had shorter cold ischemia times (6 versus 7 hours), and more often died from trauma (48% versus 32%) versus stroke (24% versus 55%). Indeed, other publications have demonstrated equivalent survival metrics when favorable donor (eg, age and warm ischemia time) and transplant parameters (eg, cold ischemia time) are carefully considered among a transplant center's DCD selection criteria.19, 20 In particular, DCD donors who were 45 years old or younger, had a warm ischemia time of 15 minutes or less, and had a cold ischemia time of 10 hours or less yielded livers that performed similarly to the livers of their DBD counterparts according to national data.19 In a separate study of the registry, donor warm ischemia times of less than 30 minutes and cold ischemia times of less than 10 hours were characteristic of low-risk DCD grafts, which led to outcomes indistinguishable from those of DBD livers.20
Additionally, in the study by Taner et al.,14 the donor warm ischemia time was relatively short (25.3 ± 10.8 minutes). Consequently, the donor warm ischemia time did not emerge as a significant predictor of either graft failure (P = 0.067) or ischemic cholangiopathy (P = 0.764) because of both the narrow distribution of this covariate and the sample size limitation. However, the asystole/cross-clamp time did predict the development of ischemic cholangiopathy (hazard ratio per minute = 1.161, 95% confidence interval = 1.021-1.321, P = 0.023) in the DCD cohort. This interval includes the mandatory waiting period (2 minutes according to the Society of Critical Care Medicine21 or 5 minutes according to the Institute of Medicine22) and the relatively short operative time necessary to initiate the aortic flush and cross-clamping. Arguably, the longer agonal phase, which in combination with the asystole/cross-clamp time constitutes the donor warm ischemia time, is most heavily implicated in the development of ischemic cholangiopathy.11, 23-25 Nonetheless, Taner et al. are the first to report that as a proxy of surgical technique, the asystole/cross-clamp time is an important mediator of ischemic cholangiopathy in DCD livers. It is thought that warm ischemia–mediated microcirculatory impairment and thrombosis of the peribiliary vascular plexus lead to an exaggerated form of ischemia/reperfusion injury. This is characterized by platelet aggregation, leukocyte adherence, the liberation of oxygen free radicals, and cellular injury, which culminate in biliary stricture formation.26
Risk predictor data derived from complementary national registry and institutional sources can be used to construct a decision analysis framework capable of better informing decisions about DCD liver utilization. The principal mediators described in the clinical literature11, 19, 20 and confirmed by the basic science literature27, 28 are warm ischemia and, to a lesser extent, cold ischemia. These mediators in turn are moderated by other risk factors (eg, donor age, weight, and cause of death), which modulate the likelihood of ischemic cholangiopathy according to their level.10, 11 The construction of such a model supports clinician and programmatic decision making, fosters transparency in the communication of risk to our patients, and identifies gaps in our knowledge that should become the focus of future scientific inquiry.
Making decisions about the optimal utilization of DCD livers is more complicated than the mere consideration of the associated donor and transplant-related risks. Not surprisingly, recipient factors also play an important role in predicting the outcomes of DCD transplantation.10, 11, 19, 20 Mateo et al.20 demonstrated that the risk stratification of recipients by age, medical condition, previous transplantation, dialysis, and renal impairment had a greater impact on graft survival than DCD donor and transplant parameters. In contrast, Lee et al.19 demonstrated a similarly negative impact on graft survival due to donor and recipient factors. The composite risk of adverse outcomes associated with DCD liver transplantation can be presented alternatively as the number needed to harm. Interestingly, the numbers needed to harm for ischemic cholangiopathy, primary nonfunction, graft failure, retransplantation, and patient mortality at 1 year are 7, 48, 10, 13, and 22, respectively.13 Clearly, the risks associated with DCD liver utilization are not prohibitive, but they should be tempered against the risk of recipient mortality on the wait list. Schaubel et al.29 established that the transplantation of livers with a high donor risk index provided a survival benefit for candidates with high Model for End-Stage Liver Disease (MELD) scores but not for candidates with low scores.29 More recently, we sought to identify subpopulations of patients who derived the greatest benefit from DCD livers.30 We found that candidates with MELD scores higher than 20 and candidates with hepatocellular carcinoma beyond the Milan criteria benefited from DCD liver transplantation in comparison with remaining on the wait list for a DBD liver.
Transplant center performance has been under growing scrutiny, particularly since the implementation of the conditions of participation issued by the Centers for Medicare and Medicaid Services.31, 32 Coupled with the threat of regulatory oversight, the realization of inferior outcomes7, 11, 12 and greater costs9, 33 has prompted a decline in DCD liver utilization.1 The standardization of DCD procurement protocols,34 careful DCD donor and recipient selection,19, 20 and support for the investigation of future effective therapies capable of improving the quality of DCD livers are necessary to improve outcomes.26 Although readers of the study by Taner et al14 should be cautioned against wholesale acceptance of these encouraging results because of the limitations in generalizability, this study may serve as a basis for rational liver allocation policy reform. The development of a MELD exception scheme to support the retransplantation of recipients of DCD livers failing because of ischemic cholangiopathy and improvements in risk adjustment to limit punitive actions against DCD liver transplant programs are necessary to preserve DCD liver utilization and allow the optimal expansion of the donor pool.