The survival benefit of deceased donor liver transplantation using MELD 3.0: A contemporary analysis

Abstract

The decision to proceed with liver transplant (LT) must account for the risk of death without LT and the likelihood of survival after, a concept known as “survival benefit.” A 2005 study suggested that deceased donor LT (DDLT) survival benefit is achieved at a MELD score of 15, a threshold that persists today. This study reassesses that threshold in the context of MELD 3.0. Data from all adults listed for primary single-organ LT in the UNOS/OPTN registry (January 1, 2021–March 31, 2023) were analyzed. Those undergoing living donor LT were excluded. Using sequential stratification, Cox regression models comparing mortality between LT recipients and waitlist candidates estimated hazard ratios across MELD 3.0 subgroups. Among 21,594 LT candidates (median MELD 3.0, 22, IQR 16–30), LT recipients had a 95% lower adjusted mortality risk than waitlisted candidates (HR 0.05, 95% CI 0.04–0.07, p < 0.001). This survival benefit varied by MELD 3.0 score, with an advantage emerging at MELD 3.0 ≥ 12. At MELD 3.0 of 12–14, DDLT reduced mortality by 46% (HR 0.54, 95% CI 0.30–0.96, p = 0.04), while no significant survival benefit was seen below MELD 3.0 of 12, largely due to high post-LT mortality. In this analysis of UNOS/OPTN data, DDLT conferred a survival benefit at MELD 3.0 scores ≥ 12, revising the historical threshold of 15. However, MELD 3.0 alone does not capture the full complexity of LT candidacy. Factors such as refractory ascites, hepatic encephalopathy, frailty, and poor quality of life should also be considered when considering LT for the individual patient.

VISUAL ABSTRACT

INTRODUCTION

The liver transplantation (LT) system prioritizes candidates based on their risk of death without a LT, employing the Model for End-Stage Liver Disease (MELD) score as the primary ranking metric. Initially designed to predict mortality following transjugular intrahepatic portosystemic shunt procedures, MELD has played a pivotal role in liver allocation policies since 2002.^[1,2] Its ability to predict 90-day mortality in patients with cirrhosis enables a risk-based allocation approach, aiming to prioritize those facing the highest risk of death without LT.^[3]

However, when making the decision to proceed with LT, transplant clinicians must balance not only the risk of death before LT, but also the risk of death after. This balance, referred to as the “LT survival benefit”, has been operationalized by comparing the life-years gained from LT against the expected survival if the patient remained on the waitlist. A landmark study by Merion et al.^[4] published 2 decades ago, highlighted the risks of early LT, noting that it could elevate mortality rates in some patients. They suggested that analyzing survival benefit by comparing transplanted patients to similar non-transplanted candidates could provide nuanced insights into the timing of LT and guide liver allocation, ensuring equitable and effective distribution of organs.^[5] Their findings showed that patients transplanted with a MELD score below 15 faced higher 1-year post-LT mortality compared with those with the same MELD scores who remained on the waitlist. Though never intended as an absolute cutoff for LT, as the MELD score does not account for other drivers of mortality (eg, portal hypertensive complications) or indications for LT (eg, quality of life), this MELD 15 threshold has become a critical benchmark for LT referral in the absence of portal hypertensive complications or quality-of-life considerations.^[6] These insights also led to policy changes, such as the Regional and National Share rules, which prioritize organs for regions with candidates scoring 15 or higher before considering those local candidates with lower scores.^[2]

Since then, risk prediction by MELD score has continued to evolve, reflecting advancements in understanding and addressing disparities in LT. MELD-Na, introduced in 2016, enhanced waitlist mortality prediction by incorporating serum sodium. In July 2023, MELD 3.0 was adopted nationally to, among other things, mitigate sex-based disparities in access to transplant. Yet, a MELD score of “15” has persisted as the general guidance for referring patients with cirrhosis without portal hypertensive complications for DDLT. This persists even though MELD, MELD-Na, and MELD 3.0 values are not interchangeable: MELD 3.0 reclassification can lead to up-categorization in 10.1% of patients and down-categorization in 3.5% of patients, as reported in the original MELD 3.0 study.^[7] In light of these updates in the MELD score and the clinical utility of the “MELD threshold” for referral for LT, we aimed to reassess the net survival benefit of DDLT using the most updated version of the MELD score, MELD 3.0.

METHODS

Study design and data source

We conducted a retrospective, observational study using the United Network for Organ Sharing/Organ Procurement and Transplantation Network (UNOS/OPTN) registry. The UNOS database includes data on all donors, waitlisted candidates, and LT recipients in the United States.

Study population

All patients (aged 18 and above) listed for primary single-organ LT in the UNOS/OPTN registry from January 1st, 2021, to March 31st, 2023, were evaluated for inclusion in this study. Data were obtained from the UNOS/OPTN registry as of March 31st, 2023. Patients were followed to a terminal outcome in the database of either death, removal from the waitlist (for various etiologies), loss to follow-up, or end of the observation period on March 31st, 2023. Those without an available MELD score at enrollment (n = 1) and Status 1 candidates (n = 580) were excluded. Candidates with a diagnosis of HCC (n = 3986) and those granted MELD exception scores for other approved conditions (n = 880) were excluded, as their MELD scores do not reflect true physiological liver dysfunction. As this study focused on deceased donor liver transplants (DDLTs), patients who underwent living donor liver transplants (LDLTs) (n = 894), split liver recipients (n = 133), patients listed for retransplant (n = 618), or multiorgan transplants (n = 1615) were also excluded. A total of 21,594 patients were therefore included in the study. A total of 841 of the 11,251 LT recipients were missing follow-up time data since LT and were therefore not included in the sequential stratification model.

Data variables

Clinical and sociodemographic data for candidates were collected at the time of waitlist registration and included the following variables: age, sex, race/ethnicity, insurance coverage, blood type, diabetes mellitus, body mass index (BMI), Karnofsky Performance Status (KPS), etiology of liver disease, ascites, and encephalopathy. Liver disease etiologies were grouped into the following common diagnostic categories: Hepatitis C (HCV), Hepatitis B (HBV), metabolic dysfunction–associated steatotic liver disease (MASLD), alcohol-associated liver disease (ALD), autoimmune etiologies (including primary biliary cirrhosis, primary sclerosing cholangitis, and autoimmune hepatitis), and other etiologies of cirrhosis (any other listing code that met inclusion criteria). We calculated the MELD 3.0 for all patients using laboratory values at all available time points between registration and their waitlist event (eg, LT, death, delisting, last follow-up). Because MELD 3.0 was implemented in the United States in July 2023, we computed the values with the available data from the MELD score (sex, albumin, INR, creatinine, sodium, and bilirubin) imputed in the database.

Analytic approach

Our analytic approach was similar to that of prior publications evaluating LT survival benefit.^[4,8] Survival times for waitlisted candidates started for all patients at the date of listing and were censored at the date of death or on removal from the waitlist. Survival times for transplant recipients started at the date of listing and were censored at the date of death or last follow-up. Patients were stratified by the following MELD 3.0 categories: 6–11, 12–14, 15–17, 18–20, 21–23, 24–26, 27–29, 30–32, 33–35, 36–39, and 40+ to provide a representation of relative mortality risk across a wide MELD 3.0 score range and based on prior literature.^[4]

Unadjusted mortality rates were calculated as the ratio of deaths per 1000 patient-years of follow-up, specific to each MELD category. Transplant mortality rates included deaths within 1 year post-transplant. The cumulative incidence function for waitlist mortality (defined as a combination of death and delisting from being too sick to transplant), with DDLT as a competing risk, was analyzed and reported by MELD subcategory. For the unadjusted mortality rates per 1000 patient-years of follow-up and the cumulative incidence function rates with DDLT as a competing risk, the MELD subgroups were defined using the last available MELD score for waiting list patients and the final pre-transplant MELD score for transplant recipients.

Sequential stratification was used to assess the survival benefit of time-dependent LT.^[8,9] This statistical approach can be viewed as a series of experiments, matching the time from listing to LT of LT recipients to waitlist candidates who have survived at least the same length of time since listing. The approach was implemented using a stratified Cox model that is stratified on these matched groups, and robust standard errors were used to account for the fact that patients could appear in more than one stratum. MELD was included as a time-updated variable in the models; specifically, at each observed LT time, the MELD score at the start of that time-point was used as the time-updated MELD value. Findings were reported as hazard ratios (HRs) comparing mortality between waitlist candidates and transplant recipients. The HR for transplant mortality risk compared with waitlist mortality risk was estimated for the entire study population, as well as for each MELD 3.0 subgroup through the use of interaction terms, both unadjusted and adjusted. For the multivariable regression, all variables associated with waitlist mortality with a p-value of <0.10 in univariable analysis were considered for inclusion in the final multivariable model, as well as variables judged clinically significant, with the final HR adjusted for age, sex, Hispanic ethnicity, diabetes, and KPS score.

A multivariable Cox model in the LT subgroup was used to assess associations of post-LT survival with recipient and donor characteristics. As all variables included in the multivariable models had minimal missing data ( <6%), complete case analysis was used.

This study was approved by our Institutional Board Review. All statistical analyses were performed using STATA v.16 (College Station, TX) and R version 4.4.1 (Vienna, Austria), and statistical significance was defined as a two-sided p-value ≤0.05.

RESULTS

Sociodemographic characteristics of LT candidates and recipients

The sociodemographic and clinical characteristics of LT candidates and recipients are summarized in Table 1. The median age of LT candidates during the study period was 55 years (IQR 46–63), with a majority being male (60%) and predominantly White (72%). More than half of the candidates (54%) had education beyond high school, and most (67%) had a KPS score of at least 50 at the time of listing. The leading cause of liver disease among candidates was ALD, followed by other etiologies and MASLD (32%, 29% respectively). Rates of ascites and hepatic encephalopathy were 42% and 16%. The median MELD 3.0 score at listing was 22 (IQR 16–30).

TABLE 1.

Sociodemographic characteristics of liver transplant candidates at listing and at transplant

Variablesa	Liver transplant candidates at listing (n = 21,594)	Liver transplant candidates at transplant (n = 11,251)
Demographics and comorbidities
Age, y	55 (46–63)	54 (45–51)
Females	40	37.3
Ethnicity
White	72	73
Black	6.0	5.9
Hispanic	17	16
Asian	3.3	3.0
Other	1.8	1.8
Insurance
Private	27	52
Public	21	41
Missing/unknown	52	7.7
Education
High school or less	38	38
Higher degree than high school	54	56
Unknown	7.9	5.8
Blood type
O	47	45
A	38	36
B	12	14
AB	3.8	5.2
BMI, kg/m2	28 (25–33)	29 (25–33)
KPS score at listing
0–40	33	51
50–70	46	37
80+	22	11
Liver disease
Etiology
HCV	5.6	4.2
HBV	1.4	1.6
ALD	32	38
MASLD	22	21
Autoimmune	10	9.4
Other	29	26
Ascites	42	53
Encephalopathy	16	20
MELD components
MELD 3.0	22 (16–30)	29 (22–35)
Creatinine, mg/dL	1.0 (0.8–1.6)	1.1 (0.8–1.7)
Sodium, mEq/L	136 (132–138)	136 (132–139)
Bilirubin, mg/dL	3.8 (1.8–10.1)	4.0 (1.8–11.5)
INR	1.6 (1.3–2.1)	1.7 (1.3–2.3)
Albumin, g/dL	3.2 (2.8–3.6)	(2.8–3.7)

^a% or median (IQR).

Abbreviations: ALD, alcohol-associated liver disease; BMI, body mass index; KPS, Karnofsky Performance Score; HBV, hepatitis B virus; HCV, hepatitis C virus; INR, International Normalized Ratio; MASLD, metabolic dysfunction–associated steatotic liver disease.

The sociodemographic and clinical characteristics of DDLT recipients were similar to those of the candidates, though recipients had lower KPS scores, with half (51%) having a KPS score of 40 or lower at the time of LT. Like candidates, ALD was the most common etiology among LT recipients. The median MELD 3.0 score at DDLT was 29 (IQR 22–35).

MELD 3.0 scores distribution

Figure 1 shows the distribution of MELD 3.0 scores at listing and at DDLT. At listing, nearly half (45%) of DDLT candidates had a MELD 3.0 score of 20 or lower, including 2081 patients (10%) with a score of 11 or lower. Only 19% of candidates had a MELD 3.0 score of 33 or higher. At the time of transplant, in contrast, 20% of LT recipients had a MELD 3.0 score at or below 20, with the majority (60%) having scores of 27 or higher.

FIGURE 1.

Distribution of MELD 3.0 scores at listing and at transplant.

Waitlist and post-transplant mortality per MELD subcategory

When examining overall mortality rates, LT candidates had a mortality rate of 372.7 deaths per 1000 person-years, with an increase in mortality rates as MELD 3.0 scores rose (Table 2). In comparison, DDLT recipients had an overall post-transplant mortality rate of 83.2 deaths per 1000 person-years during the study period, with the highest mortality rates observed at both ends of the MELD spectrum: 111.2 deaths per 1000 person-years for those with MELD scores between 6 and 11, and 116.4 deaths per 1000 person-years for those with MELD scores of 40 or higher. LT recipients with a MELD 3.0 score below 12 experienced higher post-transplant mortality rates—111.2 per 1000 person-year—compared with 82.2 deaths per 1000 person-years among those with MELD 3.0 scores of 12 or higher.

TABLE 2.

Unadjusted waiting list and transplant mortality rates by MELD 3.0 category

	Waiting list			Transplant (1 y follow-up)
MELD 3.0	Deaths	Patient-years (PY) since listing	Rate per 1000 PY	Deaths	Patient-years (PY) since transplant	Rate per 1000 PY
6–11	157	1759	89.3	11	99	111.2
12–14	107	1137	94.1	12	157	76.4
15–17	135	1155	116.9	19	341	55.6
18–20	183	1024	178.7	39	465	83.9
21–23	190	586	324.2	38	546	69.6
24–26	182	267	682.1	44	633	69.5
27–29	170	133	1280.3	52	740	70.2
30–32	217	105	2057.3	63	783	80.4
33–35	188	63	2999.5	51	642	79.5
36–39	249	64.6	3854.8	66	688	96.0
40+	710	119	5975.7	101	868	116.4
Total	2488	6675	372.7	496	5963	83.2

Survival benefit and mortality risk by MELD subcategory

The overall covariate-adjusted mortality risk for DDLT recipients was 95% lower than for LT candidates (HR 0.05, 95% CI 0.04–0.07, p < 0.001). Indeed, after adjusting for age, sex, Hispanic ethnicity, diabetes, and KPS score—all variables significant in both univariable and multivariable models—the findings strongly suggest a survival benefit in favor of DDLT (Table 3). However, this survival benefit varied across MELD subgroups. Notably, the MELD 3.0 category of 12–14 represented the threshold where the survival benefit of DDLT was statistically significant, after which the survival benefit of DDLT increased progressively with higher MELD scores (Figure 2). Specifically, DDLT recipients in the 12–14 group had a 46% lower mortality risk compared with candidates with equivalent MELD scores (HR 0.54, 95% CI 0.30–0.96, p = 0.038). In the lower MELD group (6–11), the HR was below 1, but the result was not statistically significant. A sensitivity analysis using MELD-Na yielded comparable mortality rates (Supplemental Table S1, http://links.lww.com/LVT/B47). A sensitivity analysis stratified by liver disease etiology demonstrated results largely consistent with those of the overall cohort, though slight variations in the MELD 3.0 threshold for survival benefit were observed (15–17 for ALD and MASLD; 21–23 for cholestatic/other etiologies) (Supplemental Table S2, http://links.lww.com/LVT/B47). The viral etiology subgroup could not be evaluated due to model-fitting limitations, likely related to its smaller sample size.

TABLE 3.

Covariate-adjusted mortality hazard ratios (transplant: waiting list) overall and by MELD 3.0 score category using sequential stratification

MELD 3.0	HRa,b (95% CI)	p
6–11	0.67 (0.38–1.15)	0.146
12–14	0.54 (0.30–0.96)	0.038
15–17	0.36 (0.23–0.56)	<0.001
18–20	0.35 (0.25–0.49)	<0.001
21–23	0.20 (0.14–0.28)	<0.001
24–26	0.15 (0.11–0.20)	<0.001
27–29	0.08 (0.06–0.10)	<0.001
30–32	0.06 (0.04–0.08)	<0.001
33–35	0.04 (0.04–0.06)	<0.001
36–39	0.02 (0.01–0.03)	<0.001
40+	0.009 (0.007–0.012)	<0.001
Overall	0.05 (0.04–0.07)	<0.001

^aHazard ratio for post-transplant mortality risk at transplant compared with waitlist.

^bAdjusted for age, female sex, Hispanic ethnicity, diabetes, and Karnofsky Performance Status (KPS) score.

FIGURE 2.

Comparison of overall mortality risk expressed as a hazard ratio by MELD 3.0 score for DDLT recipients compared with candidates on the waiting list using sequential stratification. Abbreviation: DDLT, deceased donor liver transplant.

Donor quality

The higher post-LT mortality and lack of statistically significant survival benefit observed in the low MELD 3.0 group may partly reflect lower donor quality. As illustrated in Table 4, recipients with lower MELD scores tended to receive organs from older donors (median age 47 years vs. 34 years for MELD 3.0 groups 6–11 vs. 40+), with higher BMI (median BMI 29 vs. 26kg/m²), a greater prevalence of diabetes (21% vs. 7%) and a higher prevalence of donors after cardiac death (21% vs. 1%). Donor BMI was also associated with post-transplant mortality in multivariable analysis (Supplemental Table S3, http://links.lww.com/LVT/B47).

TABLE 4.

Donor characteristics, by MELD subgroups

	MELD 3.0 subgroups
Donor variablesa	6–11	12–14	15–17	18–20	21–23	24–26	27–29	30–32	33–35	36–39	40+
Age	47 (37–60)	46 (33–58)	48 (33–58)	47 (34–59)	46 (33–58)	45 (33–56)	42 (31–55)	41 (30–54)	39 (29–50)	36 (28–49)	34 (25–45)
BMI	29 (24–34)	28 (24–32)	28 (24–33)	28 (24–34)	28 (25–33)	28 (24–33)	28 (24–32)	27 (24–32)	27 (23–31)	26 (23–30)	26 (23–30)
Diabetes	21	18	16	17	16	14	12	10	10	8	7
Cold ischemia time, h	6.4 (5.2–8.5)	6.3 (4.9–8.2)	5.6 (4.5–7.1)	5.7 (4.5–7.0)	5.6 (4.5–6.9)	5.5 (4.5–6.9)	5.7 (4.6–6.8)	5.8 (4.9–6.9)	5.9 (4.9–7.0)	6.0 (5.0–7.0)	5.0 (5.0–7.2)
Share type
Local	33	40	41	45	43	42	32	29	22	22	22
Regional	31	26	28	26	27	27	30	30	32	34	31
National	36	34	32	28	30	31	38	41	46	44	47
Foreign	0	0	0	1	0	0	0	0	0	0	0
DCD	21	22	23	22	17	14	9	3	2	1	1
DRI	1.7 (0.8–2.0)	1.7 (0.6–2.0)	1.7 (0.7–1.9)	1.6 (0.6–1.9)	1.6 (0.6–1.9)	1.6 (0.5–1.9)	1.5 (0.5–1.8)	1.5 (0.5–1.8)	0.7 (0.4–1.7)	0.6 (0.4–1.7)	0.6 (0.4–1.6)

^a% or median (IQR).

Abbreviations: DCD, donor after circulatory death; DRI, Donor Risk Index.

DISCUSSION

Comparing mortality on the waitlist versus mortality after DDLT within specific MELD 3.0 subgroups, we confirmed that the survival benefit of DDLT is not evenly distributed across the full spectrum of MELD 3.0 scores. Rather, we observed that the survival benefit of DDLT was achieved at MELD 3.0 scores of 12 or higher, with the magnitude of the benefit increasing as MELD 3.0 scores rose. At MELD 3.0 scores of 11 and below, our data revealed that the survival benefit after DDLT was not significantly different from the survival experienced while remaining on the waitlist.

Our study found that patients with the lowest MELD scores had post-transplant mortality rates nearly as high as those with the highest MELD scores. The current allocation system prioritizes candidates with MELD scores of 15 or higher, and the ongoing shortage of suitable donor livers has led to a growing reliance on marginal organs for low-MELD patients. Recipients of DDLT with lower MELD scores typically received organs from older donors with higher BMIs, higher prevalence of diabetes, longer cold ischemia times, and a higher prevalence of donors after cardiac death. These lower-quality livers may obscure any potential survival benefit of DDLT at lower MELD scores.^[10,11]

A substantial number of deaths while on the waiting list also occured among patients with low MELD scores. Beyond liver function alone, complications such as refractory ascites, hepatic encephalopathy, and recurrent acute cholangitis can markedly worsen prognosis and quality of life in chronic liver disease. The presence of these conditions should prompt timely referral for DDLT, and these findings highlight the need for a more refined system to capture mortality risk among low-MELD patients, ensuring that organ allocation reflects the true clinical burden. For instance, France incorporates an exception score into its allocation policy to account for these risks, based on symptom severity and significant quality-of-life impairment (https://www.agence-biomedecine.fr/IMG/pdf/french-liver-allocation-policy-handbook.pdf).

Our findings also align with previous key studies evaluating the survival benefit of DDLT, which have shown that this benefit emerges at a MELD or MELD-Na score of 12 or higher.^[12,13] Using the most contemporary cohort of liver transplant candidates and the current MELD 3.0 scoring system, we confirmed this threshold. It is also imperative to emphasize that this study does not apply to individuals seeking LDLT, as LDLT recipients were explicitly excluded. Prior research has demonstrated that the survival benefit transition point specifically for LDLT occurs at a MELD score of 11, with a substantial increase in survival and life-years gained, along with a 34% reduction in mortality compared with waitlisted patients at this score or higher.^[14] While our findings closely approximate this threshold, they pertain to a distinct population of LT candidates and recipients, identifying a survival benefit transition point for DDLT at a MELD score of 12.

Building on these considerations, this study has limitations. DDLT recipients with low MELD scores represent a distinctive group potentially affected by underlying complications that are not fully captured by MELD alone. These complications may uniquely influence their post-DDLT mortality, which could in turn alter the assessment of survival benefit in this population. Given that the study period includes a time before the implementation of MELD 3.0, patients in this analysis were assigned a recalculated MELD 3.0 score to incorporate elements from the newer scoring system. This approach has been used in other studies, including periods with differing allocation schemas.^[14] In addition, there were a relatively small number of deaths in lower MELD groups that may impact the power of our analysis, potentially limiting our ability to demonstrate DDLT survival benefit, a limitation that is inherent when analyzing data from patients at lower MELD scores.^[3,14] Indeed, we cannot confirm a survival benefit from DDLT in the lower MELD group of 6–11, nor can we rule out that the study lacked sufficient power due to the small sample size in this subgroup. As with all retrospective registry studies, potential selection bias, including survivorship bias with preferred selection of patients surviving long enough to be listed included in this study cannot be excluded. Nonetheless, this bias likely applies across the full MELD spectrum and thus should not alter the overall demonstrated DDLT benefit. Further, additional selection bias may have been introduced in patients referred for DDLT with MELD scores < 15, given prior guidelines that have advocated not referring patients with MELD scores below 15.

Finally, we do recognize that controversy persists over the existence and application of a MELD threshold to quantify net survival benefit, particularly for DDLT. The MELD score—whether in its original form, MELD-Na, or MELD 3.0—fails to fully capture a patient’s vulnerability to death and does not completely seize the morbidity burden of end-stage liver disease. Many factors and medical conditions beyond liver function alone, such as refractory ascites, hepatic encephalopathy, or recurrent episodes of acute cholangitis, contribute to poor outcomes and diminished quality of life in patients with chronic liver disease.^[15–17] These conditions should prompt referral for DDLT, irrespective of MELD thresholds. That being said, there is pragmatic value in identifying the MELD threshold at which patients may not derive a survival benefit from DDLT compared with expectant management of their chronic liver disease on the waitlist under the current liver allocation system. And that threshold, according to our analysis, is a MELD 3.0 score of 12.

Abbreviations:

ALD

alcohol-associated liver disease

BMI

body mass index

DDLT

deceased donor liver transplant

KPS

Karnofsky Performance Status

LDLT

living donor liver transplant

LT

liver transplant

MASLD

metabolic dysfunction–associated steatotic liver disease

OPTN

Organ Procurement and Transplantation Network

UNOS

United Network for Organ Sharing.

DATA AVAILABILITY STATEMENT

The data utilized in this study are available through the United Network for Organ Sharing (UNOS) Registry.