Telephone: 415-353-9321; FAX: 415-353-8709
Use of living donor liver transplantation varies with the availability of deceased donor liver transplantation
Article first published online: 25 JAN 2012
Copyright © 2011 American Association for the Study of Liver Diseases
Volume 18, Issue 2, pages 160–165, February 2012
How to Cite
Vagefi, P. A., Ascher, N. L., Freise, C. E., Dodge, J. L. and Roberts, J. P. (2012), Use of living donor liver transplantation varies with the availability of deceased donor liver transplantation. Liver Transpl, 18: 160–165. doi: 10.1002/lt.22455
- Issue published online: 25 JAN 2012
- Article first published online: 25 JAN 2012
- Accepted manuscript online: 17 OCT 2011 10:33AM EST
- Manuscript Accepted: 12 OCT 2011
- Manuscript Received: 26 JUL 2011
The demographics of patients in the United States who undergo living donor liver transplantation (LDLT) versus patients who undergo deceased donor liver transplantation (DDLT) are interesting with respect to the demographics of the donor service areas (DSAs). We examined adult recipients of primary, non–status 1 liver-only transplants from 2003 to 2009. The likelihood of undergoing LDLT was compared to the likelihood of undergoing DDLT by multivariate logistic regression. We examined the adjusted odds ratio (OR) for undergoing LDLT versus DDLT for patients with the same diagnosis and blood type after we stratified the DSAs into quintiles by the median match Model for End-Stage Liver Disease (MELD) scores. LDLT was performed for 1497 of 32,927 liver transplants (4.5%). LDLT decreased in frequency by approximately 30% from 2003 to 2009. In comparison with DDLT recipients, LDLT recipients were younger and had higher albumin levels, lower body mass indices, and lower match MELD scores. Females had increased odds of LDLT in comparison with males (OR = 1.74, P < 0.001). Patients with MELD exception scores were less likely to undergo LDLT (OR = 0.22, P < 0.001). Patients with cholestatic liver disease (adjusted OR = 2.04, P < 0.001) or malignant neoplasms other than hepatocellular carcinoma (adjusted OR = 3.33, P < 0.001) were more likely than patients with hepatitis C virus to undergo LDLT. Other characteristics associated with decreased odds of LDLT were black race (adjusted OR = 0.41, P < 0.001) and government insurance (adjusted OR = 0.51, P < 0.001). LDLT was more frequent in DSAs with high median MELD scores; the adjusted OR for LDLT was 38 for the DSAs in the highest quintile (P < 0.001). In conclusion, there are significant differences associated with race, insurance, sex, MELD exceptions, and DSA MELD scores between patients who undergo LDLT and patients who undergo DDLT. These differences can be hypothesized to be driven in part by the relative availability of LDLT versus DDLT at both the patient level and the DSA level. Liver Transpl 18:160–165, 2012. © 2011 AASLD.
Strong et al.1 reported the first successful application of living donor liver transplantation (LDLT) in 1990, and since then, living donor liver donation has emerged as the predominant source of liver allografts in Asian countries. In contrast, the application of LDLT has been guarded among those on the wait list in the United States and Europe, with LDLT accounting for approximately 3% to 4% of all liver transplants. The limited use of LDLT most likely is due to competition with deceased donor liver transplantation (DDLT) because the number of deceased donors available in the United States and Europe is larger. This competition between LDLT and DDLT is heightened by the well-publicized risks posed to the living liver donor, which may decrease the willingness of centers to undertake LDLT even when the availability of DDLT is less than adequate. In addition, a clear relationship between the volumes of LDLT centers and patient outcomes has been demonstrated, and this may limit the widespread application of LDLT.2
In this study, we sought to further define the nationwide population of LDLT recipients versus DDLT recipients. By examining the demographics of LDLT recipients, the severity and etiologies of their liver disease, and socioeconomic factors associated with LDLT use, we hoped to further our understanding of the competition between LDLT and DDLT at the patient level and the ways in which this is affected by the socioeconomic, sex, and disease-specific disparities that exist. We examined the donor service area (DSA)–specific differences in the use of LDLT versus DDLT because the absolute availability of organs to patients is determined at the DSA level.
PATIENTS AND METHODS
Data on adult, non–status 1 liver-only transplants between January 1, 2003 and December 31, 2009 were obtained from United Network for Organ Sharing Standard Transplant Analysis and Research files so that characteristics associated with LDLT could be evaluated. Only primary transplants were included. We compared the frequency distributions (categorical variables) and means (continuous variables) of patients undergoing LDLT and patients undergoing DDLT. The variables that were considered included recipient race/ethnicity, sex, blood type, primary diagnosis at the time of transplantation [categorized as hepatitis C virus (HCV), hepatitis B virus (HBV), cholestatic liver disease/cirrhosis, malignant neoplasms, and other], insurance type, diabetes, life support, preoperative portal vein thrombosis, previous abdominal surgery, Model for End-Stage Liver Disease (MELD) exception status, mechanical ventilation before transplantation, calendar year of transplantation, age, albumin level at the time of transplantation, body mass index (BMI), and match MELD score. Univariate odds ratios (ORs) for the likelihood of undergoing LDLT versus DDLT were calculated by logistic regression, and variables with a P value less than 0.1 were further evaluated by multivariate logistic regression. Stepwise selection with an α value of 0.01 was used to select the final multivariate model. The Hosmer-Lemeshow goodness-of-fit test (P = 0.19) and the area under the receiver operating characteristic curve (0.87) confirmed the model fit for the final multivariate model.
Although the P value for the MELD exception status was less than 0.001, the MELD exception status was not included in the final multivariate model because we examined the match MELD score, not the laboratory MELD score. Patients without exception scores had match MELD scores reflective of their laboratory MELD scores, whereas patients with exception points had match MELD scores greater than their laboratory MELD scores because the final exception scores were used. The inclusion of the exception status and the use of the match MELD score would have given excess weight to the patients with exception scores. Because only 9.0% of the patients undergoing LDLT were granted exception points and because the match MELD score was included in the multivariate analysis, the MELD exception score was excluded.
To evaluate the impact of the MELD score at the DSA level on the likelihood of LDLT, we calculated the median match MELD scores for DDLT for each DSA by the primary diagnosis and the blood type, which are the 2 most significant recipient factors associated with LDLT after the match MELD score (as demonstrated by the highest Wald chi-square values in the multivariate model). Each transplant recipient was assigned a DSA subgroup–specific match MELD score to represent the median match MELD score within the recipient's DSA for a DDLT of the same diagnosis and blood type. The assigned median match MELD scores were categorized into quintiles at the patient level (quintile 1, MELD score = 6-21; quintile 2, MELD score = 22; quintile 3, MELD score = 23-24; quintile 4, MELD score = 25-27; and quintile 5, MELD score ≥28): the lowest quintile represented the DSAs with the lowest subgroup-specific median match MELD scores, and the highest quintile represented the DSAs with the highest subgroup-specific median match MELD scores. We calculated the median match MELD score and the proportion of LDLT versus DDLT by the quintile of the assigned median match MELD scores. We also examined the adjusted odds of LDLT after stratification by the quintile of the assigned median match MELD scores.
Although patients who were on the liver transplant wait list for longer periods were more likely to undergo LDLT [adjusted OR for the logarithm of wait-list days = 1.18 (data not shown)], the inclusion of days on the wait list did not meaningfully change the multivariate analysis, and 41 recipients (12 with LDLT) were excluded because they spent 0 days on the wait list (a value that cannot be log-transformed). Therefore, the time on the wait list was excluded from the final model. The same was true for the transplant center volume because with the addition of MELD quintiles to the multivariate model, the transplant center volume no longer had statistical significance in the multivariate model [P = 0.66 (data not shown)]. Thus, the transplant center volume was excluded from the final model.
All statistical analyses were conducted with SAS version 9.2 (SAS Institute, Inc., Cary, NC). This study was determined to be exempt by the committee on human research of the University of California San Francisco.
There were 32,927 adult primary liver-only transplants between 2003 and 2009 that were eligible for study inclusion (Table 1). LDLT was performed for 1497 of these cases (4.5%). During the examined time period, LDLT decreased in frequency by approximately 30% (from 241 recipients in 2003 to 159 recipients in 2009), whereas DDLT increased in frequency by approximately 15% (4035 recipients in 2003 to 4615 recipients in 2009). The LDLT recipients were younger than the DDLT recipients [51 versus 54 years (mean age)] and had higher albumin levels at the time of transplantation (3.0 versus 2.9 g/dL), lower BMIs (26.7 versus 28.3 kg/m2), and lower match MELD scores (15 versus 24). In comparison with male sex, female sex was associated with increased odds of LDLT (OR = 1.74, P < 0.001). LDLT recipients were less likely to have MELD exceptions for hepatocellular carcinoma (HCC) or other reasons (OR = 0.22, P < 0.001). According to the multivariate analysis (Table 2), the likelihood of undergoing LDLT was greater for recipients with a primary liver diagnosis of cholestatic liver disease (adjusted OR = 2.04, P < 0.001) or malignant neoplasms other than HCC (adjusted OR = 3.33, P < 0.001) versus recipients with a primary diagnosis of HCV. The recipient characteristics associated with decreased odds of LDLT were blood types AB and B versus blood type O (adjusted OR for blood type AB = 0.15, P < 0.001; adjusted OR for blood type B = 0.59, P < 0.001), black race versus white race (adjusted OR = 0.41, P < 0.001), and government insurance versus private insurance (adjusted OR = 0.51, P < 0.001; Table 2).
|Characteristic||Deceased Donors [n (%)]||Living Donors [n (%)]||Univariate Analysis|
|OR (95% CI)||P Value|
|Patients||31,430 (95.5)||1497 (4.5)|
|White||23,067 (73.4)||1269 (84.8)||1.00|
|Black||2633 (8.4)||50 (3.3)||0.35 (0.26-0.46)||<0.001|
|Other||5730 (18.2)||178 (11.9)||0.57 (0.48-0.66)||<0.001|
|Female||9604 (30.6)||649 (43.4)||1.74 (1.57-1.93)||<0.001|
|O||13,978 (44.5)||706 (47.2)||1.00|
|A||11,660 (37.1)||609 (40.7)||1.03 (0.93-1.16)||0.55|
|AB||1617 (5.1)||25 (1.7)||0.31 (0.21-0.46)||<0.001|
|B||4175 (13.3)||157 (10.5)||0.75 (0.62-0.89)||0.001|
|HCV||10,515 (33.5)||415 (27.7)||1.00|
|HBV||753 (2.4)||23 (1.5)||0.77 (0.51-1.19)||0.24|
|Cholestatic liver disease/cirrhosis||2626 (8.4)||386 (25.8)||3.72 (3.22-4.31)||<0.001|
|Malignant neoplasms other than HCC||344 (1.1)||66 (4.4)||4.86 (3.67-6.44)||<0.001|
|Other||17,192 (54.7)||607 (40.6)||0.90 (0.79-1.02)||0.09|
|Private||19,240 (61.2)||1163 (77.7)||1.00|
|Government||10,740 (34.2)||305 (20.4)||0.47 (0.41-0.53)||<0.001|
|Other||1450 (4.6)||29 (1.9)||0.33 (0.23-0.48)||<0.001|
|Diabetes||6924 (22.0)||281 (18.8)||0.82 (0.72-0.93)||0.003|
|Life support||1056 (3.4)||10 (0.7)||0.19 (0.10-0.36)||<0.001|
|Portal vein thrombosis||1670 (5.3)||52 (3.5)||0.64 (0.48-0.85)||0.002|
|Previous abdominal surgery||11,647 (37.1)||575 (38.4)||1.06 (0.95-1.18)||0.29|
|Active MELD/PELD exception case||9871 (31.4)||135 (9.0)||0.22 (0.18-0.26)||<0.001|
|Mechanical ventilation||875 (2.8)||8 (0.5)||0.19 (0.10-0.38)||<0.001|
|2003||4035 (12.8)||241 (16.1)||1.00|
|2004||4378 (13.9)||266 (17.8)||1.02 (0.85-1.22)||0.85|
|2005||4540 (14.4)||252 (16.8)||0.93 (0.78-1.11)||0.43|
|2006||4738 (15.1)||219 (14.6)||0.77 (0.64-0.93)||0.007|
|2007||4571 (14.5)||191 (12.8)||0.70 (0.58-0.85)||<0.001|
|2008||4553 (14.5)||169 (11.3)||0.62 (0.51-0.76)||<0.001|
|2009||4615 (14.7)||159 (10.6)||0.58 (0.47-0.71)||<0.001|
|Characteristic||Deceased Donors (Mean ± SD)||Living Donors (Mean ± SD)||Univariate Analysis|
|OR (95% CI)||P Value|
|Age (years)||54 ± 9||51 ± 12||0.98 (0.97-0.98)||<0.001|
|Albumin level at transplantation (g/dL)||2.9 ± 0.7||3.0 ± 0.7||1.13 (1.05-1.22)||<0.001|
|BMI (kg/m2)||28.3 ± 5.7||26.7 ± 5.0||0.95 (0.94-0.96)||<0.001|
|Match MELD score||24 ± 7||15 ± 6||0.80 (0.79-0.81)||<0.001|
|OR (95% CI)||P Value|
|Cholestatic liver disease/cirrhosis||2.04 (1.72-2.42)||<0.001|
|Malignant neoplasms other than HCC||3.33 (2.32-4.78)||<0.001|
|Age (years)||0.99 (0.98-0.99)||<0.001|
|Albumin level at transplantation (g/dL)||0.78 (0.72-0.86)||<0.001|
|BMI (kg/m2)||0.96 (0.95-0.97)||<0.001|
|Match MELD score||0.79 (0.78-0.80)||<0.001|
Match MELD scores for LDLT and DDLT were compared by the assigned median match MELD scores for DDLT at the DSA by the primary diagnosis and the blood type subgroup. The sizes of the DDLT MELD quintiles varied and were as follows: 5702 patients (17%) in quintile 1, 11,121 patients (34%) in quintile 2, 5916 patients (18%) in quintile 3, 4902 patients (15%) in quintile 4, and 5286 patients (16%) in quintile 5. The median match MELD scores for DDLT ranged from 19 for DSAs in the lowest DDLT MELD quintile to 29 for DSAs in the highest DDLT MELD quintile. In contrast, the median match MELD scores for LDLT ranged from 13 for the DSAs in the lowest DDLT MELD quintile to 15 for the DSAs with the highest DDLT MELD quintile (Fig. 1). The proportion of LDLT increased with the DDLT MELD quintile (Fig. 2). Only 2.2% of the liver transplants in the lowest match DDLT MELD quintile were LDLT. In contrast, 10.3% of the liver transplants in the highest match DDLT MELD quintile were LDLT. Therefore, DSAs with high DDLT MELD scores had a greater proportion of LDLT recipients than DSAs with low DDLT MELD scores.
The odds of LDLT were stratified by the assigned match DDLT MELD quintile and were adjusted for those factors significantly associated with LDLT in the multivariate model (race, sex, blood type, primary diagnosis, insurance type, calendar year of transplantation, age, albumin level at the time of transplantation, BMI, and match MELD score). The adjusted OR for LDLT was 38 for the DSAs in the highest median match DDLT MELD quintile (P < 0.001) versus the DSAs in the lowest median match DDLT MELD quintile (Fig. 3).
Because of concerns about collinearity between the MELD quintiles and the true match MELD scores, we evaluated variance inflation factors (1.10 and 1.15, respectively) and confirmed the absence of collinearity. Nevertheless, we evaluated the adjusted ORs for DDLT MELD quintiles with and without adjustments for the true match MELD scores. Both analyses resulted in increasing ORs as the DDLT MELD quintile increased, but the magnitude of the ORs varied. The adjusted OR for the fifth quintile was 38 with the true match MELD scores and 7 without the true match MELD scores (data not shown).
Our study confirms that LDLT is still used for just a small fraction of the liver transplants performed in the United States (4.5%). Furthermore, during the investigated time period (2003-2009), the number of LDLT procedures decreased as the number of DDLT procedures increased. It is unclear whether this trend is the direct end result of a highly publicized living liver donor death in 2002,3 the institution of the MELD score (which resulted in the better allocation of deceased donor organs among patients at risk of death),4 the decreased reliance on LDLT (due to avenues for deceased donor pool expansion),5 or a combination of factors. The benefit of LDLT is a tradeoff between the risk posed to the donor and the relative benefit to the recipient of receiving an earlier transplant via LDLT instead of waiting for DDLT. If the perception of donor risk increases, the benefit of LDLT may be perceived as being lower, particularly if the relative supply of DDLT organs is perceived to be adequate.
The demographics of recipients undergoing LDLT versus recipients undergoing DDLT demonstrate that LDLT recipients are younger and thinner and have lower MELD scores. The low MELD scores likely represent a preference for using living donation in patients whose low MELD scores preclude them from contention for DDLT within their DSA.
Since the implementation of the MELD allocation system in 2002, there has been a demonstrated improvement in the equitable allocation of liver allografts with respect to race and ethnicity; however, disparities based on sex and geography appear to remain.6 The disparity in liver allocation and thus the disparity in wait-list mortality between males and females after the implementation of the MELD allocation system have been attributed to multiple factors, including the underestimation of the degree of renal dysfunction in female patients (which is due to the inclusion of the serum creatinine level rather than the glomerular filtration rate in the MELD score)7 and the smaller size of female recipients (which is a disadvantage at the time of donor/recipient organ matching). Indeed, it has been demonstrated that this size discrepancy between female and male recipients contributes to a 19% increased risk of wait-list mortality for women versus men with the same MELD scores.8 Although previous reports have demonstrated that the sex difference in DDLT recipients also exists in LDLT recipients,9 our study has demonstrated that female recipients have significantly increased odds of undergoing LDLT in comparison with male recipients. One could predict that LDLT would be a more suitable resource for a female recipient who could accommodate a partial liver allograft from a similarly sized or larger living donor versus a whole liver allograft from a larger deceased donor. Male recipients may be disadvantaged in finding a suitable donor for LDLT because their size may preclude the use of smaller female donors. The difficulty with finding deceased donor organs of adequate size and the advantage associated with finding a living donor capable of providing a graft of an adequate size seem to be reasonable explanations for the increased use of LDLT in females.
According to our study, a discrepancy appears to remain in the use of LDLT for black patients versus white patients. Although improvements in equitable allocation with respect to race and ethnicity have been demonstrated with the MELD system,6 black patients continue to be listed with higher MELD scores; this is indicative of an initial presentation with more advanced liver disease in comparison with white patients. This has prompted the consideration of the idea that racial barriers persist and prevent earlier access to health care for black patients with end-stage liver disease. Indeed, in our study, black patients had significantly higher initial laboratory MELD scores, final laboratory MELD scores, and match MELD scores in comparison with white patients [P < 0.001 (data not shown)]. Because black patients initially present with more advanced liver disease, this in itself (aside from referral patterns) may allow more expedient DDLT and possibly explains the observed discrepancy. The decrease in LDLT among black patients may also reflect donor socioeconomic factors or disease frequencies that preclude living donation.
The difference in LDLT use among patients with different blood types likely reflects need-based utilization after the assessment of the likelihood of undergoing DDLT with known waiting times for the different blood types. The other patient characteristics associated with the greater use of LDLT are more difficult to rationalize. There has been a concern that patients with cholestatic liver disease (particularly primary sclerosing cholangitis) are at a disadvantage for transplantation with DDLT, and this concern perhaps explains this population's increased likelihood of undergoing LDLT. The decreased use in patients with government insurance may represent issues related to socioeconomic status that affect the recipient or potential living donor pool because donation creates significant costs for the donors.
The inability to obtain MELD exception points increases the desirability of LDLT use. For example, patients with non-HCC hepatic malignancies who are not eligible for automatic MELD exception point upgrades appear to constitute a population benefiting from LDLT versus DDLT because these patients' laboratory MELD scores will likely be normal. Conversely, the decreased use of LDLT in patients with HCC and other MELD exceptions represents the increased use of DDLT due to the MELD exception scores. An additional advantage of LDLT versus DDLT is that LDLT may allow the transplantation of patients with significant quality of life issues associated with their liver disease but not addressed by the MELD score.
Unlike living donor kidney transplantation, LDLT as a form of therapy has no clear survival advantage over DDLT if DDLT is readily available.2 Furthermore, LDLT places a donor at significant risk. This donor risk, combined with an absence of a survival benefit in settings in which DDLT is readily available, allows DDLT in certain geographic areas to remain at a competitive advantage in comparison with LDLT. We have demonstrated that the proportion of LDLT cases within a DSA increases as the MELD quintile for DDLT increases, demonstrating that, DSAs with high MELD scores for DDLT allocation rely more on LDLT to address their wait-listed population. Indeed, this finding was also demonstrated in a report by Yeh et al.10 on the geographic inequities in DDLT access. They demonstrated that DSAs plagued with a shortage of suitable organs more frequently used LDLT.
A recent publication by Berg et al.11 highlights the issues of organ availability and the need for LDLT. They found that patients who underwent LDLT had a lower risk of death than patients without a living donor who waited for DDLT. This was true for recipients without HCC who had a MELD score either greater or less than 15. For patients with HCC and a MELD score less than 15, no benefit was found with LDLT. In contrast, patients with HCC and a laboratory MELD score greater than 15 benefited from LDLT. This finding, in conjunction with our demonstration that patients listed with MELD exception scores have a lower probability of undergoing LDLT (OR = 0.22), suggests that organ availability predominates in the decision to pursue LDLT. This suggests a different tradeoff between the donor's risk and the recipient's benefit for patients considering LDLT that must account for DDLT organ availability and the risk of wait-list death or dropout.
This study represents an observational, retrospective review with a national registry database. The assimilated data were not collected for the purpose of investigating the application of LDLT, and this represents a limitation of the study. Furthermore, the population of interest (LDLT recipients) was only a small fraction (4.5%) of the total number of liver transplant recipients in the investigated time period. The investigated population included recipients of either DDLT or LDLT and did not include those patients who were listed but did not undergo transplantation. The use of logistic regression with the generation of ORs may overestimate the magnitude of the association being measured. In addition, it is possible that there is a DSA-MELD threshold effect for the use of LDLT; that is, at a distinct transplant MELD score, the activation energy for creating an LDLT program is exceeded, and a program is created and maintained.
In summary, the use of living liver donation appears to be targeted at those patients who are disadvantaged for undergoing DDLT and/or reside in DSAs in which a high MELD score is required to undergo DDLT. The living liver donor option allows these patients whose MELD scores fall short of the high MELD score required for DDLT to undergo liver transplantation. We would expect changes in the allocation system (eg, a reduction in the benefit of HCC exception scores or other exception scores) to change the frequency of LDLT. Changes in organ distribution might also be expected to equalize the burden of LDLT with respect to geography.