MELD 3.0 leads to heterogeneous prioritization of men and women on the liver transplant waiting list

To the editor,

Over the last decade, the landscape of liver transplantation (LT) in the US has significantly changed, with alcohol-associated liver disease (ALD) now being the dominant waitlist indication. Concurrently, the performance of the current Model for End-stage Liver Disease Sodium (MELDNa) score in predicting the waitlist mortality has decreased.^[1] In this context, the small but rapidly growing subset of patients listed with alcohol-associated hepatitis (AH) has gained attention given their superior waitlist outcomes, suggesting over-prioritization in the status quo.^[2]

In 2021, Kim et al^[3] introduced the MELD 3.0 score in an effort to update MELD-based allocation to the current era and also address the long-recognized disadvantage that women face in LT access. MELD 3.0 was approved by the United Network for Organ Sharing (UNOS) in June 2022 and will replace MELDNa as the prioritization scoring system for LT in the US.^[4] Our objective was to assess the potential impact of adopting MELD 3.0 among specific subgroups.

This was a retrospective cohort study using national transplant registry data from UNOS. All initial adult LT waitlists between January 1, 2015, and December 31, 2021, were included. This start date represents the beginning of the rapid expansion of LT for ALD and AH in the US.^[2] Prior LT recipients and waiting lists for multiorgan transplants other than simultaneous liver-kidney transplants were excluded. Liver disease etiology was obtained from UNOS coding.

Native listing MELD, MELDNa, and MELD 3.0 (with albumin) scores, and the difference in the latter two scores, were calculated for each patient.^[3] All scores were capped at 40, as per the UNOS allocation policy. Scores were compared for patients listed with AH, non-AH ALD, and NASH, stratified by sex. The change in points obtained with MELD 3.0 was shown using a heatmap, by diagnosis group, sex, and MELDNa score. The discrimination of MELDNa and MELD 3.0 for AH for 90-day waitlist mortality was measured using Harrell’s c-index, obtained from univariable Cox models, as in the original MELD 3.0 study.^[3] Calibration plots of observed against expected probabilities for 90-day waitlist mortality were drawn using a Cox framework for MELDNa and MELD 3.0 scores ≥ 20 by diagnosis and sex.

Overall, 84,170 patients entered the LT waiting list between January 1, 2015, and December 31, 2021, of whom 1275 (1.5%) had AH, 28,809 (32.2%) non-AH ALD, and 17,479 (20.8%) NASH. The demographic characteristics of each group are shown in Supplemental Table 1 (http://links.lww.com/LVT/A347). Despite the adjustments of MELD 3.0, the listing score remained lower for women than men in the AH group (median: 38 vs. 40; P = 0.028), while this was higher for women in the non-AH ALD and NASH groups (median: 24 vs. 21 and 20 vs. 18, respectively; P < 0.001 for both; Table 1). In men with listing MELDNa <40, MELD 3.0 led to a median increase of +1 point for AH and a median point change of zero for non-AH ALD and NASH (P < 0.001). In women, this was +2 points for AH and NASH and +1 point for non-AH ALD (P < 0.001). The change in listing priority with MELD 3.0 was also heterogeneous by the MELDNa stratum (Figure 1). In men, a gain in listing points occurred across a broader range of MELDNa scores for AH and only at the highest scores for non-AH ALD candidates. In women, the median gain in listing points was most significant in candidates with NASH. The change in listing points was comparable for all groups with MELDNa > 35.

TABLE 1.

Calculated listing native MELD, MELDNa, and MELD 3.0 for patients with AH, non-AH ALD, and NASH, overall and stratified by sex^a

	Median (IQR)
	MELD	MELDNa	MELD 3.0
AH
Overall (n = 1275)	37 (29–40)	37 (31–40)	39 (32–40)
Male (n = 830)	37 (31–40)	38 (32–40)	40 (33–40)
Female (n = 445)	35 (27–39)	35 (30–39)	38 (31–40)
Non-AH ALD
Overall (n = 28,809)	19 (14–28)	22 (15–30)	22 (16–30)
Male (n = 20,901)	18 (13–27)	21 (15–29)	21 (15–29)
Female (n = 7908)	21 (15–29)	23 (16–31)	24 (18–33)
NASH
Overall (n = 17,479)	16 (12–22)	18 (13–24)	19 (14–25)
Male (n = 9251)	16 (12–21)	17 (13–23)	18 (13–24)
Female (n = 8228)	16 (12–21)	18 (14–25)	20 (15–27)

^aAll scores capped at 40 as per liver allocation policy.

Abbreviations: AH, alcohol-associated hepatitis; ALD, alcohol-associated liver disease; IQR, interquartile range; MELD, Model for End-stage Liver Disease; MELDNa, Model for End-stage Liver Disease Sodium.

FIGURE 1.

Heatmap demonstrating a median change in points with MELD 3.0 per MELDNa stratum for patients listed with AH (N = 1275), non-AH ALD (N = 28,809), and NASH (N = 17,479) diagnoses between 2015 and 2021. Note: Cells were shaded gray if they included <10 patients.

Among patients with listing MELDNa ≥ 30, the 90-day waitlist mortality was lowest for AH: 9.7% versus 13.4% for non-AH ALD and 16.7% for NASH (P < 0.001; Supplemental Table 2, http://links.lww.com/LVT/A348). As with MELDNa, the discrimination of MELD 3.0 was lowest for AH: c-index 0.75 versus 0.86 for non-AH ALD versus 0.84 for NASH (Supplemental Table 3, http://links.lww.com/LVT/A349). Improvements in discrimination with MELD 3.0 (vs. MELDNa) were marginal, even for women: c-index 0.80 versus 0.79 for AH, 0.86 versus 0.86 for non-AH ALD, and 0.85 versus 0.84 for NASH. The calibration of both MELDNa and MELD 3.0 was worst for AH, irrespective of sex (Supplemental Figures 1–3, http://links.lww.com/LVT/A350, http://links.lww.com/LVT/A351, http://links.lww.com/LVT/A352). For AH with scores ≥ 20, MELD 3.0 calibration appeared better than MELDNa for men and possibly worse for women (Supplemental Figure 1, http://links.lww.com/LVT/A350). A similar trend was seen for women with non-AH ALD (Supplemental Figure 2, http://links.lww.com/LVT/A351). For NASH with scores ≥ 20, MELD 3.0 calibration appeared slightly better than MELDNa for men and women (Supplemental Figure 3, http://links.lww.com/LVT/A352).

In this contemporary cohort of waitlisted candidates nationally, MELD 3.0 leads to heterogeneous changes in waitlist scores by waitlist indication and sex. In men, these are most apparent among AH patients, while in women the greatest increases in waitlist score are seen in those with AH or NASH. We acknowledge that AH accounts for a minority of the waitlisted candidates. Thus, the overall impact of such point differences on access to LT for others is unclear. Nevertheless, the issue of over-prioritization of AH candidates warrants further study in the MELD 3.0 era, given the continued expansion of LT for this indication and their superior waitlist outcomes.^[2] In support of this, we also show that the calibration of MELD 3.0 is suboptimal for AH, akin to MELDNa, with a tendency to overestimate the 90-day waitlist mortality. However, MELD 3.0 may offer small improvements in calibration over MELDNa for men with AH or non-AH ALD and for patients with NASH.

There are limitations to this study. These include the suboptimal sensitivity of the UNOS AH diagnosis code, leading to underestimation of the cohort listed with AH and misclassification of a small number of patients in the non-AH ALD group.^[5] Additional studies are needed to investigate the MELD 3.0 performance beyond the subgroups evaluated. Further research should also explore the impact of changes in priority score with MELD 3.0 on transplant rates and organ acceptance patterns using match run data.

While MELD 3.0 may help alleviate the disparity in liver allocation for women, it could lead to heterogeneous prioritization of patients with different underlying waitlist indications. Implementing a correction factor, MELD ceiling, or other approaches for conditions such as AH remains an important consideration in the MELD 3.0 era to ensure a more equitable, urgency-based allocation system for LT.

Abbreviations:

AH

alcohol-associated hepatitis

ALD

alcohol-associated liver disease

LT

liver transplantation

MELD

Model for End-stage Liver Disease

MELDNa

Model for End-stage Liver Disease Sodium

NASH

non-alcohol steatohepatitis

UNOS

United Network for Organ Sharing