- Top of page
Demand for liver transplantation continues to exceed donor organ supply. Comparing recipient survival to that of comparable candidates without a transplant can improve understanding of transplant survival benefit. Waiting list and post-transplant mortality was studied among a cohort of 12 996 adult patients placed on the waiting list between 2001 and 2003. Time-dependent Cox regression models were fitted to determine relative mortality rates for candidates and recipients. Overall, deceased donor transplant recipients had a 79% lower mortality risk than candidates (HR = 0.21; p < 0.001). At Model for End-stage Liver Disease (MELD) 18–20, mortality risk was 38% lower (p < 0.01) among recipients compared to candidates. Survival benefit increased with increasing MELD score; at the maximum score of 40, recipient mortality risk was 96% lower than that for candidates (p < 0.001). In contrast, at lower MELD scores, recipient mortality risk during the first post-transplant year was much higher than for candidates (HR = 3.64 at MELD 6–11, HR = 2.35 at MELD 12–14; both p < 0.001). Liver transplant survival benefit at 1 year is concentrated among patients at higher risk of pre-transplant death. Futile transplants among severely ill patients are not identified under current practice. With 1 year post-transplant follow-up, patients at lower risk of pre-transplant death do not have a demonstrable survival benefit from liver transplant.
- Top of page
Liver transplantation has altered the natural history of end-stage liver disease and is now considered the preferred therapy for a wide range of previously fatal chronic hepatic diseases. Optimal timing of liver transplantation in the course of disease is desirable to avoid harm from intervening too early and futility from transplanting too late. Unfortunately, little is known about the most appropriate time in the course of chronic liver disease to offer a transplant for the majority of patients, with the exception of primary biliary cirrhosis and other cholestatic disorders (1–3). Appropriate timing is especially important because of the uncertain availability of a donor organ engendered by the imbalance between the size of the pool of suitable recipients and the number of available deceased donors (4).
Given the severe shortage of donor organs, there is a need to identify which patients derive significant survival benefit from transplantation and which do not. As early as 1992, Delmonico et al. noted that assessment of outcome should be a component of allocation policy so that futile transplants could be avoided (5), and a recent report continued to support this view (6). To date, however, only pre-transplant mortality risk has been incorporated into liver allocation policy. The most recent change in liver allocation policy in the United States, by which candidates are ranked according to a more highly stratified index of pre-transplant mortality (7), continues to offer donor livers on this basis.
Comparisons of mortality risk among transplant recipients to mortality risk among listed transplant candidates with chronic end-stage renal disease, matching on time since initial listing, have been used to assess transplant survival benefit (8). This approach avoids inappropriate comparisons to the entire population of patients with end-stage disease, only a small fraction of whom are suitable candidates to be listed for transplantation. Registered transplant candidates who have already entered the transplant pathway constitute the best comparison group to transplant recipients.
In the field of renal transplantation, no distinctions are drawn among candidates based on medical urgency when allocation priority for deceased donor organs is being considered. This makes assessment of survival benefit more straightforward (8). In contrast, liver transplant allocation is based on the severity of the underlying disease necessitating hepatic replacement. Specifically, the current allocation scheme for candidates with chronic end-stage liver disease utilizes a measure of each patient's pre-transplant mortality risk that is calculated based on laboratory values of serum bilirubin, international normalized ratio of prothrombin time and serum creatinine. The score is derived from published data on the Model for End-stage Liver Disease (MELD), which predicts 3-month mortality risk for adult cirrhotics undergoing transjugular intra-hepatic portosystemic shunt procedures as well as other chronic liver disease populations and liver transplant candidates (9–11). A similar system (PELD) has been developed for pediatric candidates (12).
The tacit assumption that liver transplantation provides a large differential between waiting list mortality risk and post-transplant mortality risk for candidates at all MELD levels has not been previously tested at the national level. In this article, we use data from all liver transplant centers in the United States to examine the survival benefit attributable to liver transplantation, with a focus on adult patients with chronic liver disease at the low and high ends of the pre-transplant risk spectrum.
- Top of page
Characteristics of the study population at the time of placement on the waiting list are shown in Table 1. Just over 60% of candidates were aged 50 or older at listing. Males made up 64% of the cohort and over 87% of the candidates were white. Hispanic ethnicity was noted for 14.6% of candidates. The most common diagnosis group was non-cholestatic cirrhosis; hepatitis C cirrhosis was the most common individual diagnosis (32%).
Table 1. Characteristics of study population at wait-listing (n = 12 996)
| African American||963||7.4|
| Caucasian||11 246||86.5|
| Non-Hispanic||11 096||85.4|
| Hepatitis C cirrhosis||4189||32.2|
| Alcoholic cirrhosis||2044||15.7|
| Hepatitis B cirrhosis||534||4.1|
| Non-cholestatic cirrhosis (other)||2644||20.3|
| Primary sclerosing cholangitis||586||4.5|
| Primary biliary cirrhosis||539||4.2|
| Cholestatic cirrhosis (other)||30||0.2|
| High school or less||4914||37.8|
At the time of listing, more than one-half of candidates had a MELD score less than 15 and 9% had a score above 26 (Figure 1). The distribution of MELD scores at the time of transplant showed a shift toward higher scores (more than 25% above 26), although 24% of transplants were performed at MELD scores less than 15 (Figure 2). The proportion of transplant recipients with a MELD score at transplant less than 12 was 10%.
A total of 1538 pre-transplant deaths and 255 post-transplant deaths were observed (Table 2). The unadjusted waiting list mortality rate was 217 deaths per 1000 patient years, compared to 184 deaths per 1000 patient years among transplant recipients. There was approximately 300-fold range of waiting list mortality rates between patients with MELD scores 6–11 and those with MELD scores of 40 and higher. Patients with MELD scores 40 and higher had unadjusted post-transplant mortality rates that were more than 50% greater than that of the MELD 6–11 category.
Table 2. Unadjusted waiting list and transplant mortality rates by MELD category
|MELD||Waiting list||Transplant (1 year follow-up)|
|Deaths||Patient years (PY)||Rate per 1000 PY||Deaths||Patient years (PY)||Rate per 1000 PY|
Overall covariate-adjusted mortality risk from the Cox regression model was 79% lower for liver transplant recipients compared to candidates on the waiting list (hazard ratio = 0.21; 95% confidence interval = 0.18–0.24; p < 0.001), indicating a significant survival benefit of liver transplantation. However, the salutary effect of transplantation on survival varied across the range of MELD scores (Figure 3). With 1 year of available post-transplant follow-up, MELD category 15–17 represented a transition point. Post-transplant mortality risk was more than three times higher than waiting list mortality for MELD scores 6–11 (hazard ratio = 3.64; 95% confidence interval = 2.23–5.95; p < 0.001). The hazard ratio was also greater than 1.0 among patients with MELD 12–14 (hazard ratio = 2.35; 95% confidence interval 1.48–3.76; p < 0.001). Conversely, significant transplant survival benefit was observed at MELD scores 18 and higher, and the magnitude of transplant benefit increased with increasing MELD score (Figure 3).
Figure 3. Comparison of mortality risk expressed as hazard ratio by MELD score for recipients of liver transplants compared to candidates on the liver transplant waiting list.
Download figure to PowerPoint
The majority of patients with low MELD scores showed little to no progression over time. Among 1861 patients who were initially listed in the MELD (6–11) category and had at least 1 year of follow-up, 75.9% remained in that category 1 year later. Less than 5% progressed into a MELD score category associated with significant survival benefit.
Some patients were assigned a MELD score of 40 by the OPTN even though their calculated score was higher, because current allocation policy caps the score. Among 570 patients with a capped MELD score, 469 (82%) had an actual calculated MELD greater than 40 (range: 41–68). A separate Cox regression model for patients with MELD scores greater than or equal to 40 was fitted using the actual calculated MELD scores showed that mortality risk (and thus transplant benefit) for transplant recipients among these most ill patients remained stable (hazard ratio = 1.00; 95% confidence interval 0.89–1.11).
The observed contrast between post-transplant and waiting list mortality varied strongly across the follow-up period (Table 3). Generally, post-transplant mortality within each MELD category decreased as follow-up time increased. At MELD 6–11, post-transplant mortality was significantly higher than waiting list mortality for the first post-transplant month, while recipients transplanted at MELD scores of 30 and higher had a significant survival benefit within the first post-transplant week. Recipients with MELD scores 21–29 demonstrated significant survival benefit after the first post-transplant week.
Table 3. Covariate-adjusted mortality hazard ratios (transplant: waiting list) by MELD score category and post-transplant follow-up time
|MELD Category||Days post-transplant|
|0–7 days||8–30 days||31–365 days|
|HR*||(95% CI)†||p-value||HR*||(95% CI)†||p-value||HR*||(95% CI)†||p-value|
|6–11||33.03||(13.74, 79.22)||<0.001||9.26||(3.34, 25.69)||<0.001||1.79||(0.89, 3.59)||0.10|
|12–14||27.91||(11.81, 66.00)||<0.001||1.24||(0.17, 8.96)||0.83||1.76||(1.01, 3.06)||0.04|
|15–17||5.86||(2.57, 13.35)||<0.001||2.52||(1.23, 5.16)||0.01||0.82||(0.52, 1.28)||0.37|
|18–20||2.83||(1.04, 7.72)||0.04||1.13||(0.46, 2.78)||0.79||0.48||(0.31, 0.76)||0.002|
|21–29||1.65||(1.00, 2.73)||0.05||0.50||(0.29, 0.86)||0.01||0.15||(0.11, 0.22)||<0.001|
|30–39||0.20||(0.09, 0.43)||<0.001||0.09||(0.05, 0.17)||<0.001||0.06||(0.04, 0.08)||<0.001|
|40‡||0.11||(0.04, 0.26)||<0.001||0.04||(0.01, 0.10)||<0.001||0.02||(0.01, 0.03)||<0.001|
- Top of page
The ability of MELD to predict mortality in patients with chronic liver disease has contributed to the evolution of liver allocation policy in the United States (9,10). It is established that MELD is an important predictor of waiting list mortality (11,17,18), and several authors have recently reported that MELD is predictive of post-transplant mortality risk (6,19). One study failed to show a significant relationship between pre-transplant MELD score and post-transplant outcome, but this report was from a single-center study with limited sample size (20). In our study, we found approximately a 300-fold range of mortality risk across the MELD spectrum for patients on the waiting list, and a 1.5-fold range for post-transplant mortality risk. The availability of serial MELD data allows detailed analyses of the survival benefit of liver transplantation to be performed, comparing waiting list and post-transplant mortality risk within MELD categories.
This study demonstrates significant overall liver transplant survival benefit for patients with chronic liver disease, and provides important validation of the procedure as life-saving therapy. Deceased donor liver transplant recipients have an overall covariate-adjusted mortality risk that is less than one-quarter that of patients on the waiting list; this finding holds with and without the inclusion of ΔMELD as a covariate.
Liver transplant survival benefit is not evenly distributed across the range of MELD scores. This is particularly useful in considerations of organ allocation, where relative survival benefit among a group of candidates must be determined. On the other hand, the physician and candidate faced with a decision regarding wait-listing or acceptance of an offer of a donor liver must consider whether that particular candidate's expected lifetime with a transplant, given their current medical condition as reflected by their MELD score or other factors, is predicted to exceed that expected if the transplant is not performed. Our analyses facilitate discussion of both these issues.
At MELD scores 18 and higher, significant and progressively increasing survival benefit was demonstrated. At the highest end of the risk spectrum, there does not appear to be a MELD score above which transplantation is clearly futile. In a supplementary analysis, a Cox model fitted to the subgroup of patients with uncapped MELD scores of 40 and greater showed that the mortality hazard ratio continued to favor transplant. However, the rate of transfer to inactive status on the waiting list increases at higher MELD scores (data not shown), suggesting that effective clinical decision making and recipient selection is at work among these sick candidates. In fact, at the highest end of the MELD scale, waiting list mortality risk is extremely high, whereas post-transplant mortality risk rises much more gradually. These relationships help to explain the progressively higher benefit of liver transplantation as MELD increases.
In contrast to the findings for patients with high MELD scores, post-transplant mortality risk for the nearly one in four recipients who received liver transplants for chronic liver disease at a MELD score less than 15 was significantly higher than for comparable candidates on the waiting list. This suggests that liver transplant is more hazardous than remaining on the waiting list in such cases, based on 1 year of post-transplant follow-up. With longer follow-up, a survival benefit for low MELD patients may eventually be demonstrated, but in relative terms that benefit is likely to be of lesser magnitude than for those with higher MELD scores. Moreover, three-quarters of patients with MELD scores between 6 and 11 at listing continue to have MELD scores in that range 1 year later, suggesting that the risk of rapid progression is low.
On initial examination, it may be surprising that so many patients with low MELD scores receive liver transplants, as the current allocation policy is intended to direct donor livers to high MELD score candidates. Two related factors explain this finding. First, the MELD scores of the highest ranking candidate in each OPO vary considerably, reflecting the marked heterogeneity in the acuity of liver transplant candidates across the United States (21). Second, organs are offered to chronic liver disease candidates in descending MELD score order within the OPO service area where the donor is identified before they are offered to candidates outside that OPO. Thus, one-half of all livers are offered first to candidates with MELD scores less than 18 (data not shown). These factors collectively lead to the current situation in which over 40% of all liver transplants are given to patients with MELD scores below 18.
This study has implications for liver allocation policy in the United States. The current ordering of liver transplant candidates with chronic liver disease appears to be suboptimal. Distribution of donor livers to candidates with higher MELD scores in a larger initial geographic unit than the OPO, extending to the regional level, would simultaneously steer a higher proportion of scarce organs to patients likely to benefit from them and reduce the probability of transplant for those with less likelihood of benefit. The latter element is important, given that candidates with very low MELD scores cannot yet be shown to benefit from a transplant, and may even be harmed by one. The Board of Directors of the OPTN recently approved such a modification to the national allocation system for deceased donor livers in the United States (22).
Separately, establishment of a minimum MELD score for placement on the waiting list would be a more direct method of reducing the proportion of transplants given to patients with low MELD scores. Selection of a minimum listing MELD score needs to be informed by several considerations. With sufficiently long follow-up, a subset of patients with MELD scores below 18 may ultimately be shown to derive significant benefit. It may be imprudent, therefore, to suggest a minimum MELD score for listing that is too close to the currently calculated transplant benefit transition range of 15–17. On the downside, if patients were no longer able to be listed for transplant at low MELD scores, it might be more difficult for transplant physicians to study the course of end-stage liver disease. It has been suggested that transplant teams might not be permitted by third-party payers to follow patients referred for transplantation unless they are placed on the waiting list. These considerations must be weighed against the potential harm of doing a liver transplant that does not significantly prolong the life of the recipient, particularly as candidates with higher MELD scores and greater expected benefit are available on the waiting list. Reevaluation of individual benefit calculations will be useful when longer follow-up becomes available, although it will not likely change the ordering of benefit by MELD.
In summary, our analysis adds important data to the continuing evolution in organ allocation policy. These results suggest that transplants are being performed for some candidates who have a higher risk of dying from the transplant procedure than they have of dying from their underlying liver disease. Rather than relying solely on the risk of pre-transplant death, the survival benefit among candidates should be considered as a component of allocation policy in order to direct organs to those most likely to benefit from the procedure. The current practice of transplanting substantial number of patients at low MELD scores, with associated lack of demonstrable transplant benefit, should be carefully reconsidered.