Organ Failure and Patient Survival

A number of biological factors can influence both short-term and long-term function of transplanted solid organs. The function of the liver, kidney, heart, lung, and pancreas depends on the continuous flow of blood through them. Ischemic time refers to the amount of time that elapses when blood flow to an organ is interrupted (e.g., when the organ is removed for transplantation).

Some organs appear more sensitive to ischemic damage than others. For example, with current technology, common general practice suggests that acceptable clinical results cannot be obtained with heart grafts exposed to much more than 4 hours of ischemia. Livers have longer acceptable ischemic times, and kidneys even longer, using preservation fluids such as University of Wisconsin solution and technologies such as pulsatile perfusion.

The duration of ischemic time is highly, positively correlated with the incidence of primary nonfunction (failure to function after a transplant). A lengthy ischemic time may also impair long-term graft function. Increased donor age and other aspects of the donor’s health status, such as condition of the organ, can accentuate the impact of ischemic time on primary graft nonfunction.

Primary nonfunction refers to a situation in which the organ, after it has been transplanted, fails to function and must be replaced. For kidney graft failure, dialysis is available as a backup. For failing hearts, ventricular assist devices may be used, at least for short periods of time. With lungs and livers, no substitute is available as a therapeutic bridge. As a result, the recipient of a failed or failing transplanted lung or liver, for example, is at risk of death if he or she does not receive a replacement. However, replacement of the failed organ with a second transplant (i.e., retransplantation) means that an organ has been used that could potentially have saved the life of another individual.

Strategies that minimize the number of organs lost to primary nonfunction are essential. This goal may be accomplished by technological advances that extend maximal achievable and, in turn medically acceptable, ischemic time. Alternatively, the more immediately available approach is to minimize ischemic time. For example, it has been suggested that the rates of primary nonfunction after liver transplantation double from approximately 4 to 8 percent when cold ischemic time is extended beyond 12 hours (Ploeg et al., 1993).

Longer ischemic time is also associated with an increased rate of delayed graft function, i.e., a situation in which the graft eventually functions, but only after a prolonged period of time. Delayed graft function, in turn, is associated with longer hospital stays, a higher rate of morbidity and mortality in the recipient, and a higher rate of late graft loss.

An approximate 4.2 percent reduction in primary graft nonfunction, achieved by eliminating severely steatotic (i.e., "fatty") livers, reducing ischemic times, and using selected patients has been reported to reduce the need for retransplantation due to primary nonfunction or initial poor function (D’Allesandro et al., 1998). Extrapolating these data to the 4,000 transplants performed nationally would mean that 170 additional patients could receive a liver transplant. This compares favorably with the increase in recovered cadaveric livers of only 231 between 1997 and 1998. This example does not prove that this strategy is correct or should be universally adopted. Rather, the example illustrates how careful scrutiny of procurement and utilization practices and subsequent clinical outcomes may be used to model and then measure optimal management of a scarce human resource.

In the early days of transplantation, the optimal approach to preserve and protect the function of organs deprived of their blood flow had not been well explored. As a result, the donor and recipient had to be located very close to each other to minimize ischemic time. Methods to improve the medically acceptable ischemic time became an intense focus of research that continues. As organ preservation and technical aspects of transplantation improve, the geographic limitations for organ transport have been eased, but not totally eliminated.

The medical literature addressing the impact of cold ischemic time on outcome is expanding but is not yet sufficiently developed to state with certainty the optimal times on an organ-by-organ basis. Even the basic criteria by which viability and function are judged in laboratory-based studies are subject to scientific debate. More to the point, the number of patient and donor variables that confound the interpretation of clinical transplant results is large. Moreover, variability among transplant programs in their philosophy regarding the use of extended criteria donors and organs, as well as the role of retransplantation, significantly affects the results produced in any series.

In addition to ischemic time, several donor factors also influence graft survival. As a result of the shortage of organs for transplantation, the criteria for organ donation have been expanded to include marginal donors (i.e., extended criteria donors) for those candidates awaiting a transplant who could face death if a donor does not become available within a limited time. Donor age, health at the time of donation, and the presence of fatty change on donor liver biopsy are all representative of donor and donor organ characteristics that may influence graft survival.

The transplant team needs to have the flexibility to apply medical judgment in selecting extended criteria donors for candidate recipients with life-threatening organ failure. These decisions may relate solely to the donor source or to the recipient’s medical status, and the results of such transplantation decisions must be weighed in clinical context. As an example, approximately 50 percent of candidates for a cardiac transplant die before a donor becomes available. In this circumstance, a 10–15 percent risk of primary graft nonfunction, hypothetically, might be acceptable if the patient was medically decompensating and likely to die if no donor were available. However, the increased use of non-heartbeating donors and other extended criteria donors must be prospectively evaluated within the context of current and novel technology. The impact on total organ allocation among potential recipients must also be assessed. These analyses must be formulated in a manner that recognizes that clinical and programmatic philosophies will influence perceived differences in outcome.

As the science of organ preservation continues to advance, the duration of tolerable ischemic time from organ procurement to organ transplantation may increase. An important distinction must be made, for purposes of this analysis, between what might be labeled "maximal achievable cold ischemic time" (i.e., the longest duration of cold storage to which an ideal organ can be exposed and still have some measurable chance of functioning when reanastomosed to a blood supply) and "medically acceptable ischemic time" (i.e., the duration of cold ischemia that has been associated in clinical experience with an appropriate and acceptable percentage of acute and long-term organ survival). These times may differ significantly. Improvements in the former rely primarily on advances in technology, which are then explored in clinical studies to determine the rates of acute and long-term graft function. In addition, although the maximal achievable ischemic time may be an absolute, the medically acceptable ischemic time will differ depending on the relative scarcity of the organ, the opportunities for retransplantation, the condition of the patient, and increasing knowledge of synergistic variables that influence ultimate organ survival.

Based on a review of the existing literature on organ preservation and patient survival, outlined in Tables 6-2 through 6-6, the committee generated a summary of its findings, which are presented in Table 6-1. The figures presented in Table 6-1 are not meant to be standards of practice, but rather approximations that will vary as a function of other factors (described above). Although these findings should not be interpreted as absolute standards, they tend to agree in general with the current practice among transplant professionals.

Any strategy to expand organ allocation areas, for example, as described in this report, would have to take into account the very significant efforts devoted to matching a suitable donor with a suitable recipient, including the mechanisms currently used by the OPO system to expedite organ recovery and distribution. Given current biological constraints, any format must have as a central goal an organ allocation policy that serves to minimize ischemic time within reasonable limits in locating a potential recipient. That this function can be performed for some organs on a large geographic basis with some efficiency is attested to by current practice nationwide as well as the results within regional sharing programs.

Health outcomes data of several different types will be needed to assess and monitor the impact of biological factors on the organ distribution and allocation system. The data collected should inform the evaluation of minimum performance criteria for the organ procurement process and the transplantation process itself because they may have an impact on organ viability. Rigorous evaluation of the procurement process would appear to be a sound principle.

Data provided by the United Network of Organ Sharing (UNOS) suggest that between 450 and 550 (or 10–13 percent) of livers recovered per year (1994–1998) are not transplanted. It is difficult to ascertain the exact reasons for this, although possibilities include a marginal donor, difficulty in finding a second center in a timely manner after the first choice rejects the organ, or the finding of extensive steatosis or hepatitis in the donor organ. Each of these losses may be unavoidable. Alternatively, many of these lost opportunities might be avoided by improved communication and tracking—for example, data on the time from notification of a possible donor to the time that formal contact between the OPO and family is established; time to obtain permission for donation; time to scheduling of organ harvest; duration of the organ harvest procedure; number of organs procured but not used; and cold ischemic time of procured organs stratified by appropriate geographic criteria (e.g., miles traveled).

Transplant center-based measures would likely include the number of delivered but discarded organs; number of transplanted organs with primary nonfunction or delayed graft function; and the number of patients requiring retransplantation. Both acute and chronic organ survival could be followed and analyzed by appropriate demographics to suggest where more efficient organ allocation might be implemented to maximize organ utilization. A method to ensure the accuracy of data reporting as well as the timely availability of data is essential.

Despite the variable nature of patients and donors, other parameters that are well within the control of the system may be associated with divergent results. Appropriate and timely data analysis will strengthen the ability of the medical and allied communities to make strategic decisions in this regard. Promulgation and enforcement of minimum performance guidelines should help optimize graft survival of the overall population. Given the critical nature of this system, all involved parties should be monitored for quality control and quality assurance and for compliance with recommended methods and processes. Lastly, appropriate measures are needed to assess the impact of the Final Rule on the biological and practical measures that affect organ failure and patient survival. It must also be recognized that as methods for preservation or other technologies change, the system must be flexible enough to incorporate new data. The National Marrow Donor Program is offered as an example of a system that has operated well with respect to many of these factors (see Box 6-1).

Historically, the primary approach to exploring the impact of various changes on the allocation system has been through the use of computer simulation models. These models allow the user to input various characteristics of the organ allocation system (e.g., initial waiting list composition, recipient stream, status changes, donor stream, allocation policy, liver offer/acceptance process, post-transplant relisting/mortality) and then simulate the impact that various changes in organ allocation policies have on relevant outcomes (e.g., numbers of primary and repeat transplants; distribution of transplants by medical urgency status; posttransplant survival rates; percentage of transplants performed locally, regionally, and nationally; cost-related measures; and, waiting times).

As an illustration, change from the current allocation policy to a system using expanded allocation areas is generally expected to increase the number of status 1 patients receiving liver transplants and decrease the number of status 3 patients receiving transplants. Depending on the assumptions of the model, this change can lead to either increased or decreased posttransplant survival. The outcomes and conclusions of the simulation models are highly dependent on the assumptions upon which they are based.

This has largely been the case for the two major simulation models used in this area; the Pritsker model used by UNOS and the CONSAD model used by the University of Pittsburgh. In general, the Pritsker model shows that national organ sharing will result in more repeat transplants and poorer posttransplant survival than will the current system (Edwards and Harper, 1995). Although there is some evidence of reduced pretransplant mortality, it is at the expense of increased posttransplant mortality.

The CONSAD model also shows a decrease in pretransplant deaths but an increase in posttransplant deaths (CONSAD, 1995). The two models differ slightly because the CONSAD model assumes that, under a national sharing system, status 3 patients are at increased risk of death following transplant. The CONSAD model also shows a larger number of status 1 patients would die under a national system than does the Pritsker model.

Those developing the Pritsker model had an advantage over the CONSAD model because of their complete access to all center-level data from UNOS. Furthermore, they were able to validate their simulation model results using the rates actually observed in the population of transplant patients over time.

In an effort to better understand the determinants of organ failure and posttransplant survival, the committee examined posttransplant mortality data for liver transplant recipients who were transplanted in 1998 and 1999, using the data provided to this committee by UNOS. Attention was restricted to this more current period because of the change by UNOS in 1998 to the definitions of medical urgency status categories. This time restriction severely limits both the length of follow-up and the number of transplanted patients for which follow-up information was available. Therefore, this analysis should be replicated as more follow-up data under the new status system become available.

The sample was comprised of 1,095 transplanted patients in status categories 1, 2B, and 3. The follow-up period ranged from 0.03 month to 17.83 months, with an average follow-up of 3.30 months. The committee examined blood type (O and B versus A and AB), age (0–5, 6–17, and 18 and older), gender, race (black versus other), status (1, 2B, and 3), OPO volume (small, medium, and large),^* and follow-up time as potential predictors of patient survival. In addition, OPO-specific effects were included as a random effect in the model. A mixed-effects "person–time" logistic regression model was used to analyze these data and follows directly from the previously described mixed-effect multinomial logistic regression model, where interest is restricted to only two outcomes (i.e., dead or alive).

Results of the analysis revealed that risk of posttransplant mortality for status 1 patients significantly decreased over time ([MMLE] = –0.58, SE = 0.089, p < .001). Similarly, patients transplanted in status 2B (MMLE = –0.74, SE = 0.298, p < .01) and status 3 (MMLE = –1.37, SE = 0.529, p < .01) both had decreased risk of mortality relative to patients transplanted in status 1.

The analysis also showed that patients located in smaller-volume OPOs had increased risk of posttransplant mortality relative to those in larger-volume OPOs (MMLE = 0.79, SE = 0.323, p < .01). These results are not readily explainable. Because smaller OPOs have a larger proportion of status 2B and status 3 patients receiving transplants than larger OPOs, smaller OPOs should be expected to have lower mortality rates. The results found may be explained with the fact that, as a general rule, smaller OPOs are serving lower volume transplant centers. There is considerable health services research indicating that, for a variety of other surgical procedures, there is a positive correlation between volume and patient outcomes (Hannan, 1995; Hosenpud, 1994). Although the committee did not find comparable research for liver transplantation, it did find that the 1997 Report of Center Specific Graft and Patient Survival Rates, produced by UNOS (UNOS, 1997), contains a table showing that several of the transplant centers doing 25 or fewer liver transplants had 1-year graft survival rates significantly lower than expected, given the health status of their patients (See Fig. II-2, pg. 15, UNOS, 1997). Further research is needed before any definitive conclusion can be drawn. Therefore, the committee is reluctant to draw any inference as to whether or how graft and patient survival might be affected by the broader sharing of organs.

Ischemic times for solid organs have not been rigorously evaluated in the past and they are an important factor in the calculus of allocation. The committee reviewed existing literature and made judgments based on this information that are in general agreement with current practices. Data analysis also supports the previously reported association between volume and outcome—in this case, larger OPOs are associated with decreased mortality rates following transplantation.

Table 6-2

Table 6-2