Validation of MELD3.0 in 2 centers from different continents

Abstract

Background: MELD3.0 has been proposed to stratify patients on the liver transplant waiting list (WL) to reduce the historical disadvantage of women in accessing liver transplant. Our aim was to validate MELD3.0 in 2 unique populations.

Methods: This study is a 2-center retrospective cohort study from Toronto, Canada, and Valencia, Spain, of all adults added to the liver transplant WL between 2015 and 2019. Listing indications whose short-term survival outcome is not adequately captured by the MELD score were excluded. All patients analyzed had a minimum follow-up of 3 months after inclusion in the WL.

Results: Six hundred nineteen patients were included; 61% were male, with a mean age of 56 years. Mean MELD at inclusion was 18.00 ± 6.88, Model for End-Stage Liver Disease Sodium (MELDNa) 19.78 ± 7.00, and MELD3.0 20.25 ± 7.22. AUC to predict 90-day mortality on the WL was 0.879 (95% CI: 0.820, 0.939) for MELD, 0.921 (95% CI: 0.876, 0.967) for MELDNa, and 0.930 (95% CI: 0.888, 0.973) for MELD3.0. MELDNa and MELD3.0 were better predictors than MELD (p = 0.055 and p = 0.024, respectively), but MELD3.0 was not statistically superior to MELDNa (p = 0.144). The same was true when stratified by sex, although the difference between MELD3.0 and MELD was only significant for women (p = 0.032), while no statistical significance was found in either sex when compared with MELDNa. In women, AUC was 0.835 (95% CI: 0.744, 0.926) for MELD, 0.873 (95% CI: 0.785, 0.961) for MELDNa, and 0.886 (95% CI: 0.803, 0.970) for MELD3.0; differences for the comparison between AUC in women versus men for all 3 scores were nonsignificant. Compared to MELD, MELD3.0 was able to reclassify 146 patients (24%), the majority of whom belonged to the MELD 10-19 interval. Compared to MELDNa, it reclassified 68 patients (11%), most of them in the MELDNa 20-29 category.

Conclusions: MELD3.0 has been validated in centers with significant heterogeneity and offers the highest mortality prediction for women on the WL without disadvantaging men. However, in these cohorts, it was not superior to MELDNa.

Graphical abstract

FIGURE 1

Comparison of areas under the curve for 90-day mortality prediction on the waiting list between the different scoring systems. MELD3.0 offered the best prediction (p = 0.008). Transplanted patients in the first 90 days are excluded. Abbreviations: MELD, Model for End-Stage Liver Disease; ROC, receiver operating characteristic.

FIGURE 2

Number of patients reclassified by MELD3.0 by sex. (A) Compared to MELD. Overall. Thirty percent of women were reclassified as gaining priority when using MELD3.0 versus MELD as opposed to 19% for men (p = 0.002). (B) Compared to MELDNa. Overall. Fourteen percent of women were reclassified as gaining priority when using MELD3.0 versus MELDNa as opposed to 6% for men (p < 0.001). Red circles indicate the number of patients who gain priority. Blue dotted circle indicates the number of patients who lose priority. The main diagonal indicates patients who do not change priority. Abbreviations: MELD, Model for End-Stage Liver Disease; MELDNa, Model for End-Stage Liver Disease Sodium.

FIGURE 3

Probability density distribution of transplants for patients on the waiting list per center (p < 0.001).

FIGURE 4

Overall survival by sex, including the pretransplant and posttransplant periods.