Advertisement

Towards optimally replacing the current version of MELD

      Linked Article

      To the Editor:
      I was delighted to read Ge et al.’s review (2022)
      • Ge J.
      • Kim W.R.
      • Lai J.C.
      • Kwong A.J.
      Beyond MELD” – emerging strategies and technologies for improving mortality prediction, organ allocation and outcomes in liver transplantation.
      published in the Journal of Hepatology that included a summary of the leading liver allocation models and computational methodologies, including model for end-stage liver disease (MELD)-Plus, a model which was developed as a collaboration between Massachusetts General Hospital and IBM Research in 2017.
      • Kartoun U.
      • Corey K.
      • Simon T.
      • Zheng H.
      • Aggarwal R.
      • Ng K.
      • et al.
      The MELD-Plus: a generalizable prediction risk score in cirrhosis.
      The authors also highlighted MELD 3.0, which was proposed to replace the current version of MELD, the MELD-Na score.
      • Kim W.R.
      • Mannalithara A.
      • Heimbach J.K.
      • Kamath P.S.
      • Asrani S.K.
      • Biggins S.W.
      • et al.
      MELD 3.0: the model for end-stage liver disease updated for the modern era.
      The methodology of developing MELD 3.0 involved evaluating and using additional variables that were not included in the MELD-Na model. The performance of MELD 3.0 seems promising, especially because it correctly reclassified a net of 8.8% of decedents to a higher MELD tier, resulting in a simulated model with fewer waitlist deaths compared to MELD-Na. Although MELD 3.0 seems promising as a potential top candidate to replace MELD-Na, I would like to raise a few points for further consideration.
      The development of MELD 3.0 included the evaluation of additional variables on top of MELD-Na’s, including measurable variables such as albumin and height; however, the evaluation did not include components of other leading models, such as lactate (MELD-lactate),
      • Mahmud N.
      • Asrani S.K.
      • Kaplan D.E.
      • Ogola G.O.
      • Taddei T.H.
      • Kamath P.S.
      • et al.
      The predictive role of model for end-stage liver disease-lactate and lactate clearance for in-hospital mortality among a national cirrhosis cohort.
      blood urea nitrogen (MELD-GRAIL-Na),
      • Asrani S.K.
      • Jennings L.W.
      • Kim W.R.
      • Kamath P.S.
      • Levitsky J.
      • Nadim M.K.
      • et al.
      MELD-GRAIL-Na: glomerular filtration rate and mortality on liver transplant waiting list.
      ,
      • Asrani S.K.
      • Jennings L.W.
      • Trotter J.F.
      • Levitsky J.
      • Nadim M.K.
      • Kim W.R.
      • et al.
      A model for glomerular filtration rate assessment in liver disease (GRAIL) in the presence of renal dysfunction.
      and MELD-Plus’s white blood cells, and total cholesterol.
      • Kartoun U.
      • Corey K.
      • Simon T.
      • Zheng H.
      • Aggarwal R.
      • Ng K.
      • et al.
      The MELD-Plus: a generalizable prediction risk score in cirrhosis.
      The rationale behind excluding these variables should be discussed.
      Note that the limitation specified in Table 1,
      • Ge J.
      • Kim W.R.
      • Lai J.C.
      • Kwong A.J.
      Beyond MELD” – emerging strategies and technologies for improving mortality prediction, organ allocation and outcomes in liver transplantation.
      that MELD-Plus can only be calculated after a cirrhosis-related hospital admission is found to be incorrect. It is true that MELD-Plus was developed based on analyzing electronic health records of cirrhosis-related admissions; however, there is no limitation to using the model at any given time point, while considering the most recent value of each variable e.g.
      • Tudoroiu M.I.
      • Constantin G.
      • Pâslaru L.
      • Iacob S.
      • Gheorghe C.
      • Popescu I.
      • et al.
      The combination of serum cystatin c, urinary kidney injury molecule-1 and MELD-Plus score predicts early acute kidney injury after liver transplantation.
      ,
      • Oliveira A.M.
      • Carvão J.
      • Abreu N.
      • Pereira V.M.
      • Ladeira N.
      • Jasmins L.
      The role of BABS and MELD-Plus scores in patients with hepatic encephalopathy. Semana Digestiva (EP-090), Centro de Congressos do Algarve – Vilamoura.
      MELD-Plus has not been studied in patients exclusively registered on the liver transplant waitlist.
      • Kim W.R.
      • Mannalithara A.
      • Heimbach J.K.
      • Kamath P.S.
      • Asrani S.K.
      • Biggins S.W.
      • et al.
      MELD 3.0: the model for end-stage liver disease updated for the modern era.
      Note, however, that all patients of the MELD-Plus study had cirrhosis at the time of their admission and an average MELD score of 14.2 (6.1), only slightly lower than the value calculated in the MELD 3.0 cohort.
      • Kim W.R.
      • Mannalithara A.
      • Heimbach J.K.
      • Kamath P.S.
      • Asrani S.K.
      • Biggins S.W.
      • et al.
      MELD 3.0: the model for end-stage liver disease updated for the modern era.
      Additionally, the prevalence of liver-related comorbidities, including ascites (37.9%), hepatic encephalopathy (28.6%), hepatocellular carcinoma (4.3%), and hepatorenal syndrome (3.2%), as well as chronic conditions not directly related to the liver, such as diabetes (57.0%), hypertension (58.9%), and heart failure (36.7%), was high in the MELD-Plus study. A cohort with such a high burden of comorbid conditions may be comparable with one that exclusively contains candidates for liver transplantation.
      Note also that MELD-Plus has a version with 7 variables that has comparable performance to the 9-variable full version and does not rely on admission length of stay.
      • Kartoun U.
      • Corey K.
      • Simon T.
      • Zheng H.
      • Aggarwal R.
      • Ng K.
      • et al.
      The MELD-Plus: a generalizable prediction risk score in cirrhosis.
      , Furthermore, while MELD 3.0 was indeed found to be more discriminating compared to MELD-Na (by approximately 0.8%), MELD-Plus was consistently found to be more discriminating compared to MELD-Na in two large independent databases (by 11.4% and 16.9% in the Mass General Brigham and IBM Explorys repositories, respectively
      • Kartoun U.
      • Corey K.
      • Simon T.
      • Zheng H.
      • Aggarwal R.
      • Ng K.
      • et al.
      The MELD-Plus: a generalizable prediction risk score in cirrhosis.
      ). I therefore encourage all members of the scientific community to assess further well-validated and well-calibrated
      • Kartoun U.
      Is MELD-plus advantageous for patients with low MELD-Na scores?.
      leading hepatic risk models, as well as their individual components, as we work toward optimally replacing the current version of MELD.

      Financial support

      The author did not receive financial support.

      Conflict of interest

      The author has no conflicts of interest to disclose as described by the Journal of Hepatology.
      Please refer to the accompanying ICMJE disclosure form for further details.

      Supplementary data

      The following are the supplementary data to this article:

      References

        • Ge J.
        • Kim W.R.
        • Lai J.C.
        • Kwong A.J.
        Beyond MELD” – emerging strategies and technologies for improving mortality prediction, organ allocation and outcomes in liver transplantation.
        J Hepatol. 2022; 76: 1318-1329
        • Kartoun U.
        • Corey K.
        • Simon T.
        • Zheng H.
        • Aggarwal R.
        • Ng K.
        • et al.
        The MELD-Plus: a generalizable prediction risk score in cirrhosis.
        PLOS ONE. 2017 Oct 25; 12 (eCollection 2017)e0186301https://doi.org/10.1371/journal.pone.0186301
        • Kim W.R.
        • Mannalithara A.
        • Heimbach J.K.
        • Kamath P.S.
        • Asrani S.K.
        • Biggins S.W.
        • et al.
        MELD 3.0: the model for end-stage liver disease updated for the modern era.
        Gastroenterology. 2021 Dec; 161 (Epub 2021 Sep 3. PMID: 34481845; PMCID: PMC8608337): 1887-1895.e4https://doi.org/10.1053/j.gastro.2021.08.050
        • Mahmud N.
        • Asrani S.K.
        • Kaplan D.E.
        • Ogola G.O.
        • Taddei T.H.
        • Kamath P.S.
        • et al.
        The predictive role of model for end-stage liver disease-lactate and lactate clearance for in-hospital mortality among a national cirrhosis cohort.
        Liver Transpl. 2021 Feb; 27 (Epub 2020 Dec 9. PMID: 33025731; PMCID: PMC7880877): 177-189https://doi.org/10.1002/lt.25913
        • Asrani S.K.
        • Jennings L.W.
        • Kim W.R.
        • Kamath P.S.
        • Levitsky J.
        • Nadim M.K.
        • et al.
        MELD-GRAIL-Na: glomerular filtration rate and mortality on liver transplant waiting list.
        Hepatology. 2020; 71: 1766-1774https://doi.org/10.1002/hep.30932
        • Asrani S.K.
        • Jennings L.W.
        • Trotter J.F.
        • Levitsky J.
        • Nadim M.K.
        • Kim W.R.
        • et al.
        A model for glomerular filtration rate assessment in liver disease (GRAIL) in the presence of renal dysfunction.
        Hepatology. 2019; 69: 1219-1230https://doi.org/10.1002/hep.30321
        • Tudoroiu M.I.
        • Constantin G.
        • Pâslaru L.
        • Iacob S.
        • Gheorghe C.
        • Popescu I.
        • et al.
        The combination of serum cystatin c, urinary kidney injury molecule-1 and MELD-Plus score predicts early acute kidney injury after liver transplantation.
        Surg Gastroenterol Oncol. 2018; 23: 121-126
        • Oliveira A.M.
        • Carvão J.
        • Abreu N.
        • Pereira V.M.
        • Ladeira N.
        • Jasmins L.
        The role of BABS and MELD-Plus scores in patients with hepatic encephalopathy. Semana Digestiva (EP-090), Centro de Congressos do Algarve – Vilamoura.
        May 29 – June 1, 2019
      1. MELD-Plus source code, GitHub; https://github.com/kartoun/meld-plus.

        • Kartoun U.
        Is MELD-plus advantageous for patients with low MELD-Na scores?.
        Transplantation. 2020 Jun; 104 (PMID: 32433233): e182https://doi.org/10.1097/TP.0000000000003207