International normalized ratio (INR) does not predict periprocedural bleeding in patients with cirrhosis, according to a meta-analysis of 29 studies.
This finding should deter the common practice of delivering blood products to cirrhotic patients with an elevated INR, reported lead author Alexander J. Kovalic, MD, of Novant Forsyth Medical Center in Winston Salem, N.C., and colleagues.
“INR measurement among cirrhotic patients is important in MELD [Model for End-Stage Liver Disease] prognostication and assessment of underlying hepatic synthetic function, however the INR alone does not capture the complicated interplay of anticoagulant and procoagulant deficiencies present in cirrhotic coagulopathy,” Dr. Kovalic and colleagues wrote in Alimentary Pharmacology & Therapeutics. “Yet, the ‘correction’ of these aberrancies among peripheral coagulation tests remains common ... even in modern practice, and not uncommonly occurs in the periprocedural setting.”
According to investigators, addressing INR with blood transfusion can have a litany of negative effects. Beyond the risks faced by all patient populations, increasing blood volume in those with cirrhosis can increase portal venous pressure, thereby raising risks of portal gastropathy or variceal hemorrhage. In addition, giving plasma products to patients with cirrhotic coagulopathy may further disrupt the balance between anticoagulants and procoagulants, potentially triggering disseminated intravascular coagulation.
Dr. Kovalic and colleagues noted that the lack of correlation between peripheral coagulation tests and bleeding risk has been a longstanding subject of investigation, citing studies from as early as 1981.
To add further weight to this body of evidence, the investigators conducted a systematic review and meta-analysis involving 13,276 patients with cirrhosis who underwent various procedures between 1999 and 2019. Primary outcomes included periprocedural bleeding events and the association between preprocedural INR and periprocedural bleeding events. Secondary outcomes included mortality, quantity of blood and/or plasma products used, and relationship between preprocedural platelet count and periprocedural bleeding events.
The analysis showed that preprocedural INR was not significantly associated with periprocedural bleeding events (pooled odds ratio, 1.52; 95% confidence interval, 0.99-2.33; P = .06), a finding that held across INR threshold subgroups. Similarly, no significant difference was found between mean INR of patients who had bleeding events versus that of those who did not (pooled mean difference, 0.05; 95% CI, 0.03-0.13; P = .23).
Preprocedural platelet count was also a poor predictor of periprocedural bleeding, with a pooled odds ratio of 1.24 (95% CI, 0.55-2.77; P = .60), although the investigators noted that platelet count thresholds varied widely across studies, from 30 to 150 × 109/L. When studies were stratified by procedural bleeding risk or procedure type, subgroup effects were no longer significant. Other secondary endpoints were incalculable because of insufficient data.
“Hopefully, these findings will spark initiation of more large-scale, higher-quality studies ... to reinforce minimizing administration of fresh frozen plasma for inappropriate correction of INR, which carries a multitude of adverse effects among cirrhotic [patients],” the investigators concluded.
Dr. Stephen Caldwell
According to Stephen H. Caldwell, MD, of the University of Virginia in Charlottesville, “The present paper augments accumulating literature over the past 15 years that INR should be discarded as a measure of procedure-related bleeding risk.”
Dr. Caldwell pointed out that “bleeding in cirrhosis is usually related to portal hypertension not with impaired hemostasis, with the occasional exception of hyperfibrinolysis, which is very different from a prolonged INR.”
He went on to suggest that the present findings should dissuade clinicians from a practice that, for some, is reflexive rather than evidence based.
“It’s remarkable how many medical practices become entrenched based on hand-me-down teaching during our early training years, and remain so for many years beyond as we disperse into various medical and surgical fields,” Dr. Caldwell said. “These learned approaches to common problems can clearly persist for generations despite overwhelming evidence to the contrary that usually evolve slowly and well-insulated within subspecialties or sub-subspecialties, and hence take several generations of training to diffuse into the wider practice of medical care for common problems. These may become matters of expedience in decision-making, much like the old antibiotic conundrum of ‘no-think-a-cillin,’ as critics referred to over-use of broad spectrum antibiotics. And so it has been with the INR.”The investigators disclosed relationships with AbbVie, Eisai, Gilead, and others. Dr. Caldwell disclosed research support from Daiichi concerning the potential role of anticoagulation therapy in preventing cirrhosis progression.
SOURCE: Kovalic AJ et al. Aliment Pharmacol Ther. 2020 Sep 10. doi: 10.1111/apt.16078.