The international normalized ratio – Great for prediction of bleeding in patients taking vitamin K antagonists, useless for prediction of bleeding in patients with chronic liver disease

Manuscript Handled by: David Lillicrap

Final decision: David Lillicrap, 07 May 2022

The lack of an association between altered international normalized ratio (INR) and the risk of bleeding in patients with chronic liver disease is not novel, and studies dating back to the 1980s already demonstrated that bleeding following invasive procedures was unrelated to the prothrombin time/INR, or to the platelet count.¹ Despite this evident lack of association, which was repeatedly confirmed through the decades in different settings, the tendency to correct INR values before procedures in patients with chronic liver disease persisted in spite of the evidence that plasma supplementation was rarely efficacious and often detrimental due to volume expansion leading to abrupt increases in portal pressure.^{2, 3} Indeed, recent guidelines of the main international gastroenterological, hepatological, and hemostasis scientific societies strongly discouraged the correction of INR before invasive procedures due to the absence of a correlation between INR values and the risk of post-procedural bleeding.^4-8 However, these guidelines are not necessarily adopted in day-to-day clinical practice. Recent surveys showed that many clinicians in fact do use the INR to predict bleeding risk and that correction of the INR with the aim of reducing bleeding risk is still common.^{9, 10}

Amidst this undisputed reality, one might wonder why we should still need the results obtained in the study by Afzal and colleagues, published in this issue of the Journal of Thrombosis and Haemostasis, to put a final “nail in the coffin” of the inappropriate use of the INR to predict bleeding in patients with chronic liver disease.¹¹ As a fact, the study by Afzal and colleagues leads us back to the roots of the appropriate use of INR as a laboratory aid to determine the intensity of vitamin K antagonist (VKA) treatment and to assess the potential bleeding risk associated with VKA use, while demonstrating how its use as a tool to assess bleeding risk in patients with chronic liver disease is inappropriate.^11-14 Indeed, this study adds to the current body of knowledge showing, in a clinical setting, what has already been well documented in preclinical studies reporting how using INR to predict bleeding in patients with chronic liver disease explores just one side of the moon — it does not consider the counterbalancing decrease in anticoagulant factors synthesized by the liver, such as protein C and antithrombin.¹⁵

The study by Afzal and colleagues — carried out in a population of more than 80 000 US veterans — clearly demonstrates that increasing INR values are not associated with an increasing risk of spontaneous or procedure-related bleeding in patients with chronic liver disease; however, they do confirm this association in patients using warfarin. In patients with chronic liver disease, an INR value above 1.5 was associated with a twofold increase in the risk of bleeding compared with patients with INR values below this threshold, but with increasing INR values no further increase in bleeding risk was observed. Low albumin levels and elevated bilirubin were independent contributors to the risk of bleeding in these patients, highlighting how the increased bleeding risk observed above the 1.5 INR threshold was mainly related to features of advanced liver disease. This observation is supported by the evidence that half of the bleeds were non-variceal upper gastrointestinal bleeds, which may not be related to haemostatic failure, but could be a consequence of portal hypertension. Thus, although advancing liver disease is associated with an increased bleeding risk, there is no dose dependency between INR and risk of bleeding. Therefore, these data may indicate that increasing severity of liver disease with increasing portal pressure rather than the development of haemostatic failure increases bleeding risk. In contrast, in patients receiving warfarin there was a clear, direct, and dose-dependent correlation between INR values and risk of bleeding, and importantly the bleeding risk only clearly increased at INR values above 2.5–3.0.¹¹ These results were further confirmed after exclusion of patients with post-traumatic, surgical, and fracture hemorrhages, thus making the results of the study applicable to a wide range of clinical situations. Lastly, in propensity-score matched patients with INR values greater than 2, one unit increase in INR had no effect on the risk of hemorrhage in patients with chronic liver disease (adjusted hazard ratio 0.84, 95% confidence interval 0.68–1.04), while it increased the risk by 4% in patients on warfarin.

Noteworthy, the results of this study indirectly confirmed the lack of association between thrombocytopenia and risk of bleeding in patients with chronic liver disease: severe thrombocytopenia — defined as a platelet count <50 × 10⁹/L — was not associated with a higher risk of bleeding in these patients, while it was among those receiving warfarin. Once again, this finding emphasizes the notion that decreased platelet count in patients with chronic liver disease might be counterbalanced by compensating factors, such as increased von Willebrand factor levels.¹⁵

Interestingly enough, among patients with chronic liver disease, the results of this study showed an association between the risk of bleeding and alcohol use, anemia, and chronic kidney disease, although one of the study limitations was the absence of detailed information about the presence of acute kidney injury, or infection, that reportedly may tip off the precarious haemostatic balance of patients with chronic liver disease towards bleeding (or thrombosis). Indeed, they did confirm that other concomitant factors may shift the balance towards bleeding, as anemia may contribute to bleeding tendency by interfering with platelet adhesion, and acute kidney injury or infection may interfere with both primary and secondary hemostasis.^16-18

Another limitation of this study is the absence of information about the transfusion of blood products in an attempt to correct the INR, and of their potential effect on bleeding. However, we feel that the results of this study justify a sobering, future recommendation on this issue that should help further support the message delivered by guidelines; that is, to avoid the correction of a spontaneously altered INR in patients with chronic liver disease. In this regard, there are accumulating data showing that: (i) the INR does not predict spontaneous or procedure-related bleeding in patients with cirrhosis, (ii) the risk of procedure-related bleeding in patients with liver disease is low,¹⁹ (iii) correction of the INR by infusion of fresh frozen plasma does not improve haemostatic potential,²⁰ (iv) fresh frozen plasma may paradoxically enhance bleeding risk by increasing portal pressure,³ and (v) fresh frozen plasma may do harm by transfusion-related side effects. We thus fully support current international guidelines and feel the time has come to change clinical practice by adopting these guidelines. The choosing wisely initiative (https://www.choosingwisely.org/) aims to raise awareness on avoiding unnecessary medical tests, treatments, and procedures. The American Association for the Study of Liver Diseases has previously issued a statement on routine administration of fresh frozen plasma to correct a prolonged INR prior to abdominal paracentesis or endoscopic variceal band ligation.²¹ We feel this statement could be updated to avoid using the INR to predict bleeding risk and to avoid correction of the INR in most, if not all, clinical situations for patients with chronic liver disease.

CONFLICT OF INTEREST

None.

AUTHOR CONTRIBUTION

Dr. Giannini and Dr. Lisman drafted the manuscript.