Determining the model for end-stage liver disease with better accuracy: Neutralizing the international normalized ratio pitfalls


  • Richard A. Marlar

    Corresponding author
    1. Pathology and Laboratory Services, Oklahoma City Veterans Affairs Medical Center, Department of Pathology, University of Oklahoma Health Sciences Center, Oklahoma City, OK
    • Laboratory Services 113, Oklahoma City Veterans Affairs Medical Center, University of Oklahoma Health Sciences Center, 921 NE 13th Street, Oklahoma City, OK 73104
    Search for more papers by this author

  • Potential conflict of interest: Nothing to report.

  • See Articles on Pages 520 and 528.

The purpose of the development of the model for end-stage liver disease (MELD) score was to determine the risk for mortality in patients with cirrhosis1–3 and to consistently prioritize patients for the allocation of organs for liver transplantation.3, 4 MELD uses 3 indices, of which the most important is the international normalized ratio (INR).1–3 However, anecdotal reports and concerns of hepatologists have suggested that the INR, at least in its current form, does not precisely and consistently reflect the severity of liver disease. The INR may vary, depending on the laboratory at which it is measured. Two articles in this issue of HEPATOLOGY5, 6 demonstrate that the current INR system is not perfect when applied to patients with cirrhosis and that a modified version is a better index for calculating the MELD score.


INR, international normalized ratio; ISI, international sensitivity index; MELD, model for end-stage liver disease; PT, prothrombin time.

Although the INR, a mathematical transformation of the prothrombin time (PT), was validated only to monitor stable oral anticoagulant patients,7 most of us have used (appropriately or inappropriately) the INR as a surrogate for the PT in other clinical conditions: factor deficiencies, lupus anticoagulants, and liver disease. It is an excellent method for monitoring warfarin therapy in our transient population because different laboratories use different commercial PT reagents (thromboplastins) to calculate the PT.8, 9 The INR normalizes the influence of these reagents for warfarin therapy and predicts fairly accurately bleeding and thrombotic risk due to oral anticoagulant use.7–9

Warfarin therapy reduces the vitamin K–dependent protein functions (that is, prothrombin, factors VII, IX, and X, and proteins C and S) not by decreasing the synthesis of the protein itself but rather by inhibiting posttranslational modification of an additional carboxyl group to the 8-10 glutamic acid residues in the amino terminal region of these proenzyme proteins.10 These modified amino acid residues allow the binding of the vitamin K–dependent protein to the phospholipid surface of the cell, thus providing assembly and orientation of the proteins for activation of the coagulation pathway.10 The INR works well for monitoring anticoagulation therapy because the warfarin mechanism is a “pure” reduction in protein function.11 On the other hand, liver disease is not so “pure”: not only does it have the component of decreased vitamin K–dependent protein posttranslational modifications, but it also has decreased synthesis of the vitamin K–dependent proteins and other coagulation proteins (factor V and fibrinogen) and increased levels of activated coagulation factors and enzymes.12 On the basis of these other abnormalities associated with liver disease, the INR system, which was created to monitor warfarin therapy, should not be expected to provide similar results between different PT reagents or therefore the same MELD scores between institutions.

A major mathematical component of the INR is the international sensitivity index (ISI), a value that defines its responsiveness and relationship to the reference standard. This value is determined by the manufacturer for each of the over 25 different commercially available thromboplastins under a strict protocol.8 Thromboplastins are from a variety of animal or recombinant sources, and this makes a patient's INR result variable in response to the different reagents. The ISI is determined on the basis of the World Health Organization reference thromboplastin with normal individuals and patients on stabilized oral anticoagulant therapy.7–9 Currently, the protocol used to determine the ISI does not test samples from patients with liver disease.

The authors of these 2 groundbreaking articles in this issue of HEPATOLOGY5, 6 address the question of the accuracy of the INR and the ISI in patients with liver disease. Both studies demonstrate that the manufacturer's ISI (based on oral anticoagulant therapy) is not accurate when patients with liver disease are being assessed. These studies went one step further and calculated ISI values with blood samples from patients with liver disease instead of samples from patients on oral anticoagulant therapy. In both studies, after the recalculation of the ISI, the MELD scores were more reproducible in patients with liver disease and significantly different from the values determined with the ISI value published by the manufacturer. As an example, in the study by Tripodi et al.,6 the difference between the standard ISI and the ISI calculated for liver disease ranged from 6% to more than 50%, depending on the reagent. These differences in the ISI correspond to clinically more accurate INR values. The use of the standard ISI does not provide a true measure of the INR in patients with liver disease. Hence, the INR values for patients with liver disease can be very different from laboratory to laboratory, from instrument to instrument, and from reagent to reagent; however, the INR does not vary for patients on stable oral anticoagulant therapy. This demonstrates why a MELD score at one hospital could be high, elevating the patient to the top of the waiting list for liver transplantation, but when the patient is transferred to the transplant center and the MELD is redetermined, the patient's MELD score is lower, and the patient is at a lower priority for liver transplantation. As the person responsible for coagulation testing, I have received a number of calls from perplexed hepatologists asking why there is a difference in the INRs between institutions when the INR was meant to be an equalizer. These studies are the explanation for this variability of the INR values and should be the basis for reconsidering the use of a different INR method for the determination of the MELD score.

On the basis of the studies presented in these 2 articles,5, 6 the coagulation community is going to have to consider changing the way in which we think about the INR (and ISI) for all the other uses of the PT. For patients with liver disease, the ISI used for oral anticoagulant therapy is inappropriate for the assessment of the severity of liver disease. To correctly assess patients with liver disease, a laboratory will have to devise a new ISI value for its PT reagents or perhaps a new INR equation to better align with the liver disease process. To accomplish this new direction, the coagulation community and hepatologists will have to work together to approach this in a manner that is simple and feasible yet still truly assesses a patient's liver disease status. Because the old PT and the current INR will not work for liver disease, we must develop another value, possibly a “liver normalized ratio” to accurately evaluate the severity of liver disease and assess patients, especially those patients on the waiting list for liver transplantation. This will mean that each PT reagent will have 2 underlying values for the sensitivity of the reagent: 1 for oral anticoagulant therapy usage and 1 for the evaluation of the severity of liver disease. It will take time to develop the correct methodology or formula and thus the value for accurate INR results in patients with liver disease. Above all else, it will take cooperation between the laboratory providing the assessment biomarkers for liver disease and the clinicians to use these new methods and/or values appropriately. The INR value should not be removed from the MELD formula but rather should be reconfigured to better reflect the complex process of liver disease. This does not have to be a long process, but we must start the development now, using these 2 seminal articles as a starting point.