274 CLINICAL AND SYSTEMATIC REVIEWS

CME

Changing Concepts of Cirrhotic Coagulopathy

REVIEW

Armando Tripodi, PhD1, 2, Massimo Primignani, MD2, 3, Pier M. Mannucci, MD2 and Stephen H. Caldwell, MD4

The state of clinical art of the coagulopathy of cirrhosis changed considerably over the last decade. Until 2005,
cirrhosis was considered as the epitome of the hemorrhagic coagulopathies and the abnormal hemostasis tests
associated with the disease were corrected with infusion of fresh frozen plasma or platelets to minimize the risk
of bleeding. Since that time, a great deal of work has been done and there is now a change of paradigm. The
prothrombin time once considered as an isolated measure of bleeding risk was rejected, and cirrhosis shifted from a
purely hemorrhagic construct to a mixed and thrombosis-prone paradigm. In this article we examine the interesting
history of how these conceptual changes came about.
Am J Gastroenterol 2017; 112:274–281; doi:10.1038/ajg.2016.498; published online 1 November 2016

The state of the clinical art regarding coagulation issues in liver prevailing idea was that the PT test was a good predictor of the
disease patients in 1999 was generally different from the concepts bleeding risk in cirrhosis. Therefore, shortening of the prolonged
that have slowly taken hold since that time when J.H. Joist, writing PT with FFP or PCC was considered as evidence that the coagu-
in this journal, introduced the term “AICF” (accelerated intravas- lopathy of cirrhosis was controlled and that the risk of bleeding
cular coagulation and fibrinolysis) and asserted that the complex could be efficaciously handled with these transfusional products.
coagulopathy of liver disease may also involve a “hypercoagula- For many years, FFP and PCC infusions were very common in
ble state” (1). Since that time, a great deal of work has slowly, but patients with cirrhosis when PT values exceeded the upper limit
surely, carried this field forward. Although the advances are broad of the normal range (usually with an arbitrary cutoff point), in
and much more translational work is needed in many areas, the spite of the fact that there were a number of clinical and labora-
single most notable clinical change is rejection of the prothrom- tory observations showing that the PT prolongation was not a
bin time (PT)-based international normalized ratio (INR) as an good predictor of bleeding after surgery or invasive procedures
isolated measure of bleeding risk in cirrhosis, and the use of arbi- (3–10). There was in fact an apparent paradox that did escape the
trary cutoff values to guide administration of procoagulants— attention of clinicians for many years: patients with near-normal
most notably fresh frozen plasma (FFP). As presaged by Joist’s PT did bleed, whereas patients with relatively abnormal PT com-
editorial (1) and confirmed in subsequent laboratory study, the monly did not. Furthermore, in vivo studies carried out by Ewe (3)
pathophysiology involved in cirrhotic coagulopathy is much more in 1981 had shown that a number of bleeding indices, including
nuanced than previously appreciated and requires a more sophis- the PT measured in the peripheral blood failed to correlate with
ticated approach to discern dominant pathways. Below, we exam- the degree of bleeding measured directly on the organ during lapa-
ine the interesting history of how these conceptual changes came roscopic liver biopsy.
about over the past 17 years since Joist’s prescient editorial (1). During the same period other scientists devoted their attention
to platelet number and function. Although cirrhosis has been rec-
ognized for a long time to be associated with thrombocytopenia
THE PAST AS PROLOGUE—CIRRHOSIS AS THE (Figure 1), Laffi et al. (11–17) showed that not only the platelet
EPITOME OF THE HEMORRHAGIC DISEASES? numbers but also their function was impaired. This knowledge was
The group of scientists working at the Angelo Bianchi Bonomi often translated into clinical practice, as shown by the common
Hemophilia and Thrombosis Center in Milano, Italy, have been practice of infusing platelets to patients with chronic liver disease
interested in the problem of hemostasis in chronic liver dis- before biopsy or other invasive procedures, particularly when
ease since the 1970’s. Mannucci et al. (2) were among the first counts were lower than 50×109/l. At that time, the skin bleeding
to explore the management of patients with cirrhosis with FFP time was considered as an index of a defective role of platelets in
or prothrombin complex concentrates (PCC). At that time the primary hemostasis. With this background, Burroughs et al. (18)

1
Department of Clinical Sciences and Community Health, Angelo Bianchi Bonomi Hemophilia and Thrombosis Center, Università degli Studi di Milano, Milano,
Italy; 2IRCCS Cà Granda Maggiore Hospital Foundation, Milano, Italy; 3Division of Gastroenterology and Hepatology, Milano, Italy; 4Division of Gastroenterology &
Hepatology, University of Virginia, Charlottesville, Virginia, USA. Correspondence: Armando Tripodi, PhD, Department of Clinical Sciences and Community Health,
Angelo Bianchi Bonomi Hemophilia and Thrombosis Center, Università degli Studi di Milano Via Pace 9, Milano 20122, Italy. E-mail: armando.tripodi@unimi.it
Received 21 April 2016; accepted 13 September 2016

The American Journal of GASTROENTEROLOGY VOLUME 112 | FEBRUARY 2017 www.nature.com/ajg

 Changing Concepts of Cirrhotic Coagulopathy 275

and Mannucci et al. (19) aimed to see whether or not desmopres- C and protein S (Figures 1–2). With this background, the newly
sin (DDAVP), that had been developed by Mannucci et al. (20) for developed TGA appeared a suitable tool to better understand the
the treatment of hemophilia and von Willebrand’s disease, was able coagulation system in patients with cirrhosis, because the degree

REVIEW
to shorten the prolonged bleeding time in cirrhosis. DDAVP was of thrombin formation and disappearance is influenced by both
indeed able to shorten the bleeding time, but this shortening did procoagulants and anticoagulants (Figure 2).
not translate into any clinical efficacy (18,19). The team carried out a case–control study on thrombin gen-
Therefore, it became clear that this test was not a good predictor eration with results expressed as endogenous thrombin potential
of bleeding in cirrhosis, and that DDAVP was not useful to pre- (ETP), which represents the total amount of thrombin that a given
vent or stop bleeding in this condition. Interestingly, de Franchis plasma sample can generate after triggering coagulation in vitro
et al. (21) had shown in 1993 that DDAVP actually worsened the by means of small amounts of tissue factor. These assay conditions
outcome in patients with variceal bleeding. Furthermore, Arshad are, at least in principle, much closer to coagulation as it occurs in
et al. (22) showed that DDVAVP infusion does not improve pri- vivo than the PT, not only because they are sensitive to the action
mary hemostasis in patients with cirrhosis. Later studies showed of pro and anticoagulants, but also because tissue factor and phos-
that the prolonged bleeding time in cirrhosis did not predict bleed- pholipids are added at much smaller concentrations than in the
ing from esophageal varices (23). Boberg et al. (24) found that PT, therefore more closely reproducing coagulation in vivo (25,26).
patients with cirrhosis and a prolonged bleeding time had lower
hemoglobin levels after liver biopsy but with no overt clinical
bleeding. All in all, these observations indicated that the test was A BREAKTHROUGH—MEASURING THE EFFECTS OF
not a good predictor of the bleeding risk. ACQUIRED PROTEIN C DEFICIENCY
The first experiment was, however, disappointing because the
ETP in patients with cirrhosis was lower than in healthy subjects.
SHIFTING THE OLD PARADIGM—THE THROMBIN- It seemed therefore that the ETP measured under these assay con-
GENERATION TEST ditions gave a response similar to that of the PT, failing to provide
In 2004 the Milano team started to become engaged in stud- evidence for the hypothesis under investigation. However, one of
ies based upon the recently developed thrombin-generation the members of the team (A.T.) realized that the ETP as carried
assay (TGA) (25,26). One of the most promising areas was the out in the first experiment was still inadequate to mimic thrombin
application of TGA to elucidate coagulation mechanisms in generation and its inhibition in vivo. Protein C (one of the most
acquired clinical conditions characterized by an increased risk of important naturally occurring anticoagulants) needs to be acti-
thrombosis or hemorrhage. Why cirrhosis? Talking with the gas- vated to express its full anticoagulant activity.
troenterologists of the Ca’ Granda Maggiore Hospital, the team Protein C activation in vivo is optimized by the presence of the
became cognizant that the prolongation of the PT (and related endothelial receptor thrombomodulin (Figure 2) that, however, is
tests), although useful to diagnose congenital hemorrhagic absent or present in only tiny amounts in plasma. It was reasoned
coagulopathies (i.e., hemophilia and allied disorders), were hardly that protein C, markedly reduced in cirrhosis (Figure 1), could not
useful for the clinical management of acquired coagulopathies contribute to the balance between pro and anticoagulants operat-
such as cirrhosis. The hypothesis was that these tests are not use- ing in vivo unless it was fully activated. There was, however, little
ful because they are indeed sensitive to the reduced liver synthesis experience on TGA with the addition of soluble thrombomodulin.
of procoagulant factors (Figures 1–2), but are insensitive to the AT very soon started in vitro experiments meant to find the appro-
parallel reduced synthesis (and reduction in plasma) of naturally priate TGA conditions through the addition of soluble thrombo-
occurring anticoagulant proteins, such as antithrombin, protein modulin. With this addition, plasma from patients with cirrhosis

Platelet–vessel Thrombin Fibrin

wall interaction generation dissolution

VWF AT, PC, PS, TFPI PAI

Pro-hemostatic
drivers
ADAMTS-13 FVIII Plasminogen

tPA
Anti-hemostatic
drivers Platelets XI, IX, X, VII, II, V, I
TAFI, anti-Plasmin

Figure 1. How hemostatic drivers change in different phases of hemostasis in patients with cirrhosis. ADAMTS-13, disintegrin and metalloprotease with
thrombospondin type 1 motif 13 which governs VWF; AT, antithrombin; PAI, plasminogen activator inhibitor; PC, protein C; PS, protein S; TAFI, thrombin
activatable fibrinolysis inhibitor; TFPI, tissue factor pathway inhibitor; tPA, tissue plasminogen activator; VWF, von Willebrand factor; VIII, XI, IX, X, VII, II, V
and I (fibrinogen) are procoagulant factors.

© 2017 by the American College of Gastroenterology The American Journal of GASTROENTEROLOGY

 276 Tripodi et al.

Procoagulant drivers Anticoagulant drivers

REVIEW

Tissue factor/VIIa TFPI

X IX XIa AT

IXa
VIIIa PC-PS-TM
Va
Xa

II

Thrombin

Figure 2. Coagulation balance.Thrombin generation is driven by procoagulant factors (roman numerals) and contrasted by anticoagulant factors. The bal-
ance is robust and stable in the healthy because of normal levels of pro and anticoagulant factors, but is unstable in cirrhosis because of the relative defi-
ciencies of the above drivers. Accordingly, it may tip toward hemorrhage or thrombosis (not uncommon in cirrhosis) depending on the circumstantial risk
factors experienced by patients during lifetime. AT, antithrombin; PC, protein C; PS, protein S; TFPI, tissue factor pathway inhibitor; TM, thrombomodulin.

did indeed generate as much thrombin as those of healthy sub- Met initially with a significant degree of skepticism, these results
jects (27). These results were interpreted as evidence that under were eventually published (27) along with an editorial (34). The
appropriate assay conditions (when both pro and anticoagulants editorial acknowledged some potential problems with PT-INR as
are allowed to have their contrasting roles), coagulation is “rebal- a measure of hemostasis but pointed out the validated use of the
anced” to normal because of the concomitant deficiency of both INR in prognostic scores such as the Child–Pugh (35) and MELD
coagulation drivers in cirrhosis (27) (Figures 1–2). (Model for end-stage liver disease) (36,37). With some reserva-
These observations, which revealed that cirrhosis patients tions, regarding reproducibility of the test based on derivation as a
with prolonged PT may actually be relatively hypercoagulable in measure of vitamin K antagonist (VKA) therapy (see below), few
terms of clot formation, have important practical implications. contest the latter point even today.
First, they explain why the PT or the PT-derived INR as stan-
dalone assays are not good predictors of bleeding in cirrhosis.
Second, they provide indirect evidence to explain why the infu- MORE ON THE PLATELET ROLE IN CIRRHOSIS
sion of recombinant activated factor VII, a potent pro-hemostatic Platelets have a dual role in hemostasis. First, they stick to the sub-
agent, was poorly effective in controlling bleeding from varices endothelium at the site of vessel injury and aggregate to initiate
(28,29) or during hepatectomy (30,31). The above findings also thrombus formation. Second, they support thrombin generation
point to the poor biological plausibility of using FFP as prophy- by assembling on their surface vitamin K-dependent coagulation
laxis or rescue therapy of bleeding, a common practice despite factors needed for thrombin generation. Adhesion and aggre-
the absence of evidence from randomized clinical trials. Thus, gation are made possible by the contribution of the multimeric
the infusion of FFP based on abnormal PT values is not justi- adhesive protein von Willebrand factor (VWF). The role of plate-
fied if one considers that thrombin generation is normal (or even lets in the setting of the TGA was then addressed by a second
higher than normal) in patients with cirrhosis. Furthermore, FFP set of experiments by the Milano team (38), which showed that
infusion is likely to result in fluid overload, which may increase platelets from cirrhotics were qualitatively able to support ade-
portal hypertension and paradoxically increase the risk of bleed- quate thrombin generation in vitro, provided their count was in
ing (32). Finally, in a recent study Tripodi et al. (33) have shown the range of 50–60×109/l (38).
that when FFP is added in vitro to plasma from patients with The other important function of platelets (adhesion mediated
cirrhosis in an amount able to mimic that achieved in vivo by by VWF), was investigated in 2006 in a seminal study authored
the infusion of 15 ml/kg, FFP lowers the PT as expected but fails by Lisman et al. (39). They assessed the ability of plasma from
to cause improvement in the physiologically more important cirrhotics to support adhesion of normal or patient platelets
thrombin generation. under flow conditions that mimic more closely the conditions

The American Journal of GASTROENTEROLOGY VOLUME 112 | FEBRUARY 2017 www.nature.com/ajg

 Changing Concepts of Cirrhotic Coagulopathy 277

operating in vivo than platelet aggregation tests. The VWF anti- venous thromboembolic disease in cirrhosis has been shown in US
gen was strongly increased in plasma from patients with cirrhosis and Danish studies by Northup et al. (47) and Søgaard et al. (48),
(Figure 1). In contrast, the collagen-binding capacity of VWF, as and portal vein thrombosis (resulting from slower flow, a proco-

REVIEW
well as the ristocetin cofactor activity, were lower in patients than agulant imbalance and vessel wall abnormalities, i.e., Virchow’s
in controls, indicating a dysfunction of VWF in supporting platelet triad) has emerged as a still controversial but sometimes signifi-
function. Patients also had less VWF multimers of high molecu- cant clinical development (49). More convincingly, Villa et al.
lar weight as well as reduced levels of the VWF cleaving protease (50) have recently reported in a non-blinded study that cirrhosis
ADAMTS-13 (Figure 1). However, adhesion of normal or patient patients treated with prophylactic doses of low molecular weight
platelets to a collagen surface was substantially increased when heparin avoided portal vein thrombosis and importantly had
these platelets were resuspended in plasma from patients with cir- fewer decompensating events and better survival with no excess
rhosis, as compared with control plasma. These results indicated of bleeding compared with untreated controls. The occurrence
that markedly elevated levels of VWF (a rather constant feature of a procoagulant imbalance was also postulated to be involved
in cirrhosis; Figure 1) contribute to the induction of primary in the progression of liver fibrosis (46,51) in a process described
hemostasis and compensate for the thrombocytopenia of cirrhot- long ago as parenchymal extinction. This association remains to be
ics. More recently, a critical review of the literature showed that completely explored as to whether it is driven by actual thrombus
the observations of platelet function defects in earlier studies, were formation or activation of hepatic stellate cells by thrombin. More
more a myth than reality (40). recently, also non-alcoholic fatty liver disease, a highly prevalent
Cumulatively, these findings were instrumental in changing condition affecting >20% of the general population, was associated
the old paradigm of thrombocytopathy/thrombocytopenia in cir- with biochemical signs of a procoagulant imbalance that is perhaps
rhosis that prevailed in the past and had prompted clinicians to responsible for the increased risk of cardiovascular disease associ-
use prophylactic platelet transfusion rather indiscriminately in ated with this condition (52). Much of this work was summarized
thrombocytopenic patients with cirrhosis prior to invasive pro- in an often cited landmark paper published in the New England
cedures. A note of caution was further provided by recent stud- Journal of Medicine on Mechanisms of Disease from the Milano
ies on the management of thrombocytopenic cirrhotics with an team in 2011 (53). The old paradigm of cirrhosis as the epitome of
agonist of the thrombopoietin receptors (eltrombopag). The lat- the acquired hemorrhagic coagulopathies was markedly changing.
ter did indeed increase platelet count but the study was prema-
turely interrupted because of excess thrombotic events in treated
patients (41). THE FALLACY OF THE INR IN CIRRHOSIS
The INR is a scale of values used for many years to express results
for the PT in patients on anticoagulant therapy with VKA. During
THE PARADOX—A PROCOAGULANT IMBALANCE IN the Charlottesville meeting in 2005 (44), it became apparent that
PATIENTS WITH CIRRHOSIS the INR when applied to patients with cirrhosis was not a good
The aforementioned findings provide evidence that consistent method to harmonize results across laboratories (54), and to cal-
and predictable hemostasis impairment of cirrhosis is more of a culate the MELD score, considered as an objective and consist-
myth than a fact, and that hemostasis is fragile but rebalanced in ent laboratory score to prioritize patients for liver transplantation
“stable” patients. However, in 2008 the Milano team showed that (37). The INR was none the less used as a system of results har-
plasma from patients with cirrhosis is characterized by a proco- monization in cirrhosis, in spite of the fact that it was devised and
agulant imbalance because of resistance to the in vitro anticoagu- validated only for patient on VKA. The key issue is that PT results
lant action of thrombomodulin (42). This resistance is related to are expressed as INR based on the equation: “INR=(PTpatient/
high plasma levels of factor VIII (procoagulant driver) contrast- PTnormal)ISI where the ISI stands for the “International Sensitiv-
ing with low protein C (anticoagulant driver), typical features of ity Index”. The latter is a measure of the PT responsiveness to the
patients with cirrhosis (42) (Figure 1). Meanwhile, other investi- coagulation defect induced by VKA (a typical characteristic of any
gators subsequently confirmed this finding in the frame of inde- given thromboplastin) relatively to the international thrombo-
pendent studies (43,44). plastin standard (55). However, it was not understood that the ISI
The causes and consequences of the procoagulant imbalance in is experimentally derived from testing plasma from patients on
plasma from patients with cirrhosis have been addressed in a series VKA, so that the regular INR is unable to normalize PT results in
of multidisciplinary meetings and symposia beginning in 2005 patients other than those taking VKA. The Milano team worked
with the 1st International Symposium on Coagulation in Liver for a few months on the issue and came to the conclusion that
Disease (www.coagulationinliverdisease.org) at Charlottesville, a new system of INR calibration could be devised by replacing
VA (45) and including subsequent meetings of this group held in plasma from patients on VKA with plasma from patients with
Groningen, London and Padua, and other symposia held as part cirrhosis, with the goal to derive an ISI valid for cirrhosis. The
of the American Association for the Study of Liver Disease or the hypothesis was that the newly developed ISI (called “ISIliver” as
European Association for the Study of the Liver (46) meetings. opposed to the “ISIvka”) could be potentially used to harmonize
A number of publications have demonstrated the fallacy of PT results in patients with cirrhosis, with the new scale called
cirrhotic “auto-anticoagulation”. The significance of peripheral “INRliver” according to the equation (56): “INRliver=(PTpatient/

© 2017 by the American College of Gastroenterology The American Journal of GASTROENTEROLOGY

 278 Tripodi et al.

PTnormal)ISIliver”. Use of unmodified and VKA-based sensitivity anticoagulants, and platelets, it is not as stable as in healthy sub-
index of different commercially available thromboplastins as indi- jects and may therefore tip toward bleeding or thrombosis,
ces in cirrhosis patients accounts for inter-laboratory variation in depending on the prevailing circumstantial risk factors (Figure 2).
REVIEW

measuring PT and INR (54,57). Joist (1) some years ago described the process as “AICF”, similar
As it happens in research, it was later realized that another group to but distinct from classical disseminated intravascular coagula-
was independently working on the very same idea (58). Curiously tion. It is possible that, as yet, little investigated aspects of cirrhotic
enough, the two manuscripts, submitted during the same days, coagulopathy hold a therapeutic key such as the presence of abnor-
were both accepted for publication in Hepatology (56,58) and mal forms of fibrinogen (dysfibrinogen), elevated fibrin degrada-
appeared in the same issue of the Journal accompanied by an edito- tion products, and decreased factor XIII which cross-links fibrin
rial written by Marlar (59). The Milano team is currently working (Greenberg, personal communication).
on this subject to see whether or not the modified calibration sys- Fibrinolysis is a mechanism that regulates digestion of fibrin
tem can be implemented in practice to calculate the MELD score. clots. Although hyperfibrinolysis has been advocated as a cause
of bleeding in cirrhosis it remains somewhat controversial. It
is estimated to be evident clinically and by laboratory testing in
CLINICAL PERSPECTIVES around 5–10% of patients and may respond to anti-fibrinolytic
The clinical approach to coagulation in liver disease has clearly therapy (61,62). However, Lisman et al. (63) reported that the
evolved away from past unfounded tenets toward a more nuanced decreased levels of the profibrinolytic drivers (Figure 1) that
appreciation of the complexity of the process and a much greater occur in cirrhosis are counterbalanced by a parallel decrease of
appreciation of the hypercoagulable aspects of cirrhotic coagu- the anti-fibrinolytic drivers. On the other hand, this mechanism
lopathy. Published work recounted in the foregoing has tremen- was not confirmed in a study by Colucci et al. (64), who showed
dously contributed to this evolution. that reduced levels of the thrombin activatable fibrinolysis inhibi-
tor is consistent with hyperfibrinolysis in cirrhosis. Furthermore,
increasing fibrinolytic capacity has been clearly demonstrated
WHAT ABOUT ALL OF THAT BLEEDING? in progressively more severe liver disease (65). This work along
Bleeding remains the dominant clinical issue, whereas inappro- with the sometimes striking response to anti-fibrinolytic therapy
priate clotting is common but often more subtle (32,60). Seared (personal observation by S.H.C.) in cases of diffuse mucosal or
into the memory of many physicians and care-providers is the refractory puncture wound bleeding lend support to hyperfi-
patient with progressive liver failure and bleeding, sometimes brinolysis as an underlying factor in some cases of cirrhosis-related
diffuse, and often with prolonged PT/INR and low-range plate- bleeding. Further studies with more standardized global assays for
lets. Past tenets held that one must then “correct” the INR and fibrinolysis are clearly needed to further resolve the issue.
platelets. Although the latter still hold some foundation (see
above), the former clearly does not. Indeed, intervention with
plasma fails to improve thrombin production in this setting PROPHYLAXIS, TREATMENT, AND TESTING FOR
and may exacerbate portal hypertension making a bad situation PATIENTS WITH CIRRHOSIS
worse (32). Until new information from clinical trials becomes available, the
Indeed, the derangements in hemostasis in cirrhosis can be following strategy based on expert opinion, albeit with little sup-
very challenging to fathom. Although hemostasis in cirrhosis porting clinical data is advisable (Table 1). Patients with stable
is rebalanced, because of the relative deficit of procoagulants, cirrhosis who undergo surgery or invasive procedures should not

Table 1. Prophylaxis and treatment

Clinical situation Intervention

Surgery or invasive procedure in patients with platelets >50×109/l.a Consider No prophylaxis with FFP, PCC, or platelet transfusion. Consider replacing
measuring fibrinogen level fibrinogen if low (<120–150 mg/dl)
Surgery or invasive procedure in patients with platelets <50×109/l.a Consider Consider platelet transfusion. Treatment with TPO agonists carries risk of
measuring fibrinogen level thrombosis but remains under clinical study with newer agents. Consider
replacing fibrinogen if low (<120–150 mg/dl)
Patients with active bleeding, especially if life threatening. Consider measuring Consider treatment with PCC. Treatment with rFVIIa is not advised.
fibrinogen level Consider replacing fibrinogen if low (<120–150 mg/dl)
In all of these situations, assess for evidence of renal failure and ongoing Treat renal failure and control infection if present
infection as both can significantly alter hemostasis
FFP, fresh frozen plasma; PCC, prothrombin complex concentrates; rFVIIa, recombinant activated factor VII; TPO, thrombopoietin.
These suggestions are based on expert opinion, not substantiated by data from clinical trials.
a
Irrespective of the prothrombin time and INR.

The American Journal of GASTROENTEROLOGY VOLUME 112 | FEBRUARY 2017 www.nature.com/ajg

 Changing Concepts of Cirrhotic Coagulopathy 279

receive prophylaxis with FFP or coagulation factor concentrates thromboelastography parameters would result in a significant
(PCC or rFVIIa) based on the PT value or the PT-derived INR. reduction of transfusion. However, the study did not incorporate
Fibrinogen levels may be replaced only if there is active bleeding a cohort of patients in whom no transfusion (platelets or FFP) was

REVIEW
or in a high-risk setting to a level of 120–150 mg/dl using cryopre- given prior to procedures. Furthermore, the only significant bleed-
cipitate. Volume restriction should be practiced due to the adverse ing event observed in the study was hemoperitoneum following
effect of volume expansion on portal pressure (32). Platelet trans- paracentesis (67), which could have been caused by mechanical
fusion should be sought if platelet counts are less than 50×109/l. injury irrespective of transfusion.
Treatment with agonists of the thrombopoietin receptor (eltrom- Global assays like thromboelastometry/thromboelastography
bopag and others) is not generally advisable at this time because are still emerging and clinicians using these tests often face a learn-
of the relative risk of thrombotic events, although further studies ing curve on their interpretation. Although sometimes a helpful
are in progress. Notably, treating renal failure and active infection adjunct for example in assessing suspected hyperfibrinolysis, these
are very important considerations in assessing hemostatic mecha- tests need a more thorough evaluation with sound study design
nisms in these patients (66). and examination of specific parameters in cirrhosis before specific
Rescue therapy should be sought in patients with active bleeding and generalizable recommendations can be made to guide therapy
(especially if life threatening) preferably by means of PCC in order (Table 2).
to avoid fluid overload and to restore both pro and anticoagulant
factors. It should be noted that there are several types of PCC and
their availability is variable and these agents are not very exten- FUTURE DIRECTIONS
sively studied in cirrhosis patients. Obviously, the risk/benefit ratio The past 17 years have seen a remarkable revolution in our under-
of such treatment should, however, be carefully evaluated in indi- standing of cirrhotic coagulopathy, especially from work at the
vidual patients as the risk of thrombotic events may be relevant. laboratory bench but also from early clinical translational studies.
In general, hematology expertise should be sought to guide these Current challenges include refinements in how coagulation mech-
interventions. anisms and bleeding risk in liver disease is measured (whether
Concerning the tests to be used for evaluating the risk of bleed- by TGAs, thromboelastography/thromboelastometry—all of
ing in patients with cirrhosis, it should be appreciated that the PT which need further prospective translational studies) (67,68) or
test and its congener, the INR, are not reliable tools and should by some other as yet to be described algorithm integrating labora-
only be used, although with some caveats (see above the issue of tory measurements with clinical signs. Therapeutic interventions
the INRliver vs the INRvka), as prognostic values. Thromboelastom- whether as prophylaxis or as rescue for active bleeding, need to
etry/thromboelastography that assess the viscoelastic properties of be examined in light of these more refined conceptual changes. In
clotting blood (i.e., the time of clot formation, the velocity and the stable patients, the administration of “anti”coagulant therapy may
maximal clot formation), although theoretically more promising actually preserve liver function. This remarkable finding could
as lab tools than the PT are not yet thoroughly evaluated in cir- have a substantial impact on the demand for liver transplantation
rhosis, except for liver transplantation, where they proved reliable and clearly warrants further study especially with newer anti-
to assist intraoperative transfusion therapy. They have never been coagulant agents.
shown to predict bleeding in cirrhosis. A recent study by De Pietri After more than a decade of work and establishment of a solid
et al. (67) investigated the role of thromboelastography as a lab scientific basis, the clinical challenges presented by these changes
tool to guide blood product transfusion in patients with cirrhosis in our understanding of cirrhotic coagulopathy are immense
prior to invasive procedures. The study concluded that using and much more complex than the old, arbitrary PT-INR cutoff.

Table 2. Clinical laboratory testing to evaluate the bleeding risk

Test Current state of the art

Prothrombin time and INR Unreliable as indicators of bleeding risk

Platelet count Reliable, although not substantiated by data from clinical trials (no accurate threshold levels are available)
Skin bleeding time Unreliable
Fibrinogen Commonly used with target of >120 mg/dl or >150 mg/dl. Understandable rational (needed for fibrin forma-
tion) but lacking good clinical trials
Thromboelastometry/thromboelastography Promising, but require further evaluation. Learning curve is evident for test interpretation
Thrombin-generation assays Very useful research tool. Promising in clinical setting but require further evaluation
Fibrinolysis (tests for hyperfibrinolysis) Promising, but standardized and reliable global assays are not widely available and require further evaluation.
Responsiveness for hyperfibrinolysis detection may vary between different global assays
INR, international normalized ratio.

© 2017 by the American College of Gastroenterology The American Journal of GASTROENTEROLOGY

 280 Tripodi et al.

Translating these advances into meaningful clinical outcomes is 15. Laffi G, Marra F, Gresele P et al. Evidence for a storage pool defect in plate-
lets from cirrhotic patients with defective aggregation. Gastroenterology
the new frontier. 1992;103:641–6.
16. Laffi G, Marra F, Failli P et al. Defective signal transduction in platelets
REVIEW

CONFLICT OF INTEREST from cirrhotics is associated with increased cyclic nucleotides. Gastroenter-
ology 1993;105:148–56.
Guarantor of the article: Armando Tripodi, PhD.
17. Laffi G, Cinotti S, Filimberti E et al. Defective aggregation in cirrhosis is
Specific author contributions: Conceived and designed the review, independent of in vivo platelet activation. J Hepatol 1996;24:436–43.
conducted literature reviews, and lead author in drafting the manu- 18. Burroughs AK, Matthews K, Qadiri M et al. Desmopressin and bleeding
time in patients with cirrhosis. Br Med J 1985;291:1377–81.
script: Armando Tripodi; editing and final approval of manuscript:
19. Mannucci PM, Vicente V, Vianello L et al. Controlled trial of desmopressin
Stephen H. Caldwell, Massimo Primignani, and Pier M. Mannucci. in liver cirrhosis and other conditions associated with a prolonged bleeding
Financial support: None. time. Blood 1986;67:1148–53.
20. Mannucci PM, Ruggeri ZM, Pareti FI et al. 1-Deamino-8-d-arginine vaso-
Potential competing interests: None.
pressin: a new pharmacological approach to the management of haemo-
philia and von Willebrands’ diseases. Lancet 1977;1:869–72.
21. de Franchis R, Arcidiacono PG, Carpinelli L et al. Randomized controlled
Study Highlights trial of desmopressin plus terlipressin vs. terlipressin alone for the treat-
ment of acute variceal hemorrhage in cirrhotic patients: a multicenter,
double-blind study. New Italian Endoscopic Club. New Italian Endoscopic
WHAT IS CURRENT KNOWLEDGE Club Hepatology 1993;18:1102–7.
✓ Cirrhosis as a hemorrhagic disease. 22. Arshad F, Stoof SC, Leebeek FW et al. Infusion of DDAVP does not improve
✓ Prothrombin time (PT) test as an index of bleeding risk. primary hemostasis in patients with cirrhosis. Liver Int 2015;35:1809–15.

✓ Fresh frozen plasma (FFP) used to correct the abnormal

23. Basili S, Ferro D, Leo R et al. Bleeding time does not predict gastrointestinal
bleeding in patients with cirrhosis. The CALC Group. Coagulation Abnor-
PT. malities in Liver Cirrhosis. J Hepatol 1996;24:574–80.
24. Boberg KM, Brosstad F, Egeland T et al. Is a prolonged bleeding time as-
WHAT IS NEW HERE
✓ Hemostasis in cirrhosis is rebalanced.
sociated with an increased risk of hemorrhage after liver biopsy? Thromb
Haemost 1999;81:378–81.
✓ Rejection of the PT test as an index of bleeding risk. 25. Hemker HC, Giesen P, Al Dieri R et al. Calibrated automated throm-

✓ No biological plausibility of correcting the PT with FFP. bin generation measurement in clotting plasma. Pathophysiol Haemost
Thromb 2003;33:4–15.
26. Chantarangkul V, Clerici M, Bressi A et al. Thrombin generation assessed
as endogenous thrombin potential (ETP) in patients with hypo- or hyper-
coagulability. Effects of phospholipids, tissue factor and residual platelets
on the measurement performed in platelet-poor and platelet-rich plasma.
REFERENCES Haematologica 2003;88:547–54.
1. Joist JH. AICF and DIC in liver cirrhosis: expressions of a hypercoagulable 27. Tripodi A, Salerno F, Chantarangkul V et al. Evidence of normal thrombin
state. Am J Gastroenterol 1999;94:2801–3. generation in cirrhosis despite abnormal conventional coagulation tests.
2. Mannucci PM, Franchi F, Dioguardi N. Correction of abnormal coagula- Hepatology 2005;41:553–8.
tion in chronic liver disease by combined use of fresh-frozen plasma and 28. Bosch J, Thabut D, Bendtsen F et al. European Study Group on rFVIIa in
prothrombin complex concentrates. Lancet 1976;2:542–5. UGI Haemorrhage. Recombinant factor VIIa for upper gastrointestinal
3. Ewe K. Bleeding after liver biopsy does not correlate with indices of bleeding in patients with cirrhosis: a randomized, double-blind trial.
periphera coagulation. Dig Dis Sci 1981;26:388–93. Gastroenterology 2004;127:1123–30.
4. McGill DB, Rakela J, Zinsmeister AR et al. A 21-year experience with 29. Bosch J, Thabut D, Albillos A et al. International Study Group on rFVIIa
major hemorrhage after percutaneous liver biopsy. Gastroenterology in UGI Hemorrhage. Recombinant factor VIIa for variceal bleeding in
1990;99:1396–400. patients with advanced cirrhosis: a randomized, controlled trial.
5. Diaz LK, Teruya J. Liver biopsy. N Engl J Med 2001;344:2030. Hepatology 2008;47:1604–14.
6. Terjung B, Lemnitzer I, Dumoulin FL et al. Bleeding complications 30. Lodge JP, Jonas S, Jones RM et al. rFVIIa OLT Study Group. Efficacy and
after percutaneous liver biopsy: an analysis of risk factors. Digestion safety of repeated perioperative doses of recombinant factor VIIa in liver
2003;67:138–45. transplantation. Liver Transpl 2005;11:973–9.
7. Grabau CM, Crago SF, Hoff LK et al. Performance standards for therapeutic 31. Planinsic RM, van der Meer J, Testa G et al. Safety and efficacy of a single
abdominal paracentesis. Hepatology 2004;40:484–8. bolus administration of recombinant factor VIIa in liver transplantation
8. Segal JB, Dzik WH. Paucity of studies to support that abnormal coagula- due to chronic liver disease. Liver Transpl 2005;11:895–900.
tion test results predict bleeding in the setting of invasive procedures: an 32. Giannini ED, Stravitz RT, Caldewell SH. Correction of hemostatic
evidence-based review. Transfusion 2005;45:1413–25. abnormalities and portal pressure variations in patients with cirrhosis.
9. Boks AL, Brommer EJ, Schalm SW et al. Hemostasis and fibrinolysis Hepatology 2014;60:1442.
in severe liver failure and their relation to hemorrhage. Hepatology 33. Tripodi A, Chantarangkul V, Primignani M et al. Thrombin generation in
1986;6:79–86. plasma from patients with cirrhosis supplemented with normal plasma.
10. Vieira da Rocha EC, D’Amico EA, Caldwell SH et al. A prospective study of considerations on the efficacy of treatment with fresh frozen plasma.
conventional and expanded coagulation indices in predicting ulcer bleeding Intern Emerg Med 2012;7:139–44.
after variceal band ligation. Clin Gastroenterol Hepatol 2009;7:988–93. 34. Bosch J, Reverter JC. The coagulopathy of cirrhosis: myth or reality?
11. Laffi G, La Villa G, Pinzani M et al. Altered renal and platelet arachidonic Hepatology 2005;41:434–5.
acid metabolism in cirrhosis. Gastroenterology 1986;90:274–82. 35. Pugh RN, Murray-Lyon IM, Dawson JL et al. Transaction of the oesopha-
12. Laffi G, Cominelli F, La Villa G et al. Reduced platelet thromboxane A2 gus for bleeding oesophageal varices. Br J Surg 1973;60:646–9.
production as a possible cause of defective platelet aggregation in cirrhosis. 36. Malinchoc M, Kamath PS, Gordon FD et al. A model to predict poor
Adv Prostaglandin Thromboxane Leukot Res 1987;17A:366–9. survival in patients undergoing transjugular intrahepatic portosystemic
13. Laffi G, Cominelli F, Ruggiero M et al. Molecular mechanism underlying shunts. Hepatology 2000;31:864–71.
impaired platelet responsiveness in liver cirrhosis. FEBS Lett 1987;220: 37. Wiesner R, Edwards E, Freeman R et al. United Network for Organ Sharing
217–9. Liver Disease Severity Score Committee. Model for end-stage liver disease
14. Laffi G, Cominelli F, Ruggiero M et al. Altered platelet function in cirrhosis (MELD) and allocation of donor liver. Gastroenterology 2003;124:91–6.
of the liver: impairment of inositol lipid and arachidonic acid metabolism 38. Tripodi A, Primignani M, Chantarangkul V et al. Thrombin generation in
in response to agonists. Hepatology 1988;8:1620–6. patients with cirrhosis: the role of platelets. Hepatology 2006;44:440–5.

The American Journal of GASTROENTEROLOGY VOLUME 112 | FEBRUARY 2017 www.nature.com/ajg

 Changing Concepts of Cirrhotic Coagulopathy 281

39. Lisman T, Bongers TN, Adelmeijer J et al. Elevated levels of von Willebrand 55. van den Besselaar AMHP, Poller L, Tripodi A. Guidelines for thromboplas-
Factor in cirrhosis support platelet adhesion despite reduced functional tins and plasmas used to control oral anticoagulant therapy. 1999;WHO
capacity. Hepatology 2006;44:53–61. Tech Rep Ser 1999;889:64–93.
40. Violi F, Basili S, Raparelli V et al. Patients with liver cirrhosis suffer from 56. Tripodi A, Chantarangkul V, Primignani M et al. The international normal-
primary haemostatic defects? Fact or fiction? J Hepatol 2011;55:1415–27. ized ratio calibrated for cirrhosis (INRliver) normalizes prothrombin

REVIEW
41. Afdhal NH, Giannini EG, Tayyab G et al. ELEVATE Study Group. Eltrom- time results for model for end-stage liver disease calculation. Hepatology
bopag before procedures in patients with cirrhosis and thrombocytopenia. 2007;46:520–7.
N Engl J Med 2012;367:716–24. 57. Porte RJ, Lisman T, Tripodi A et al. Coagulation in Liver Disease Study
42. Tripodi A, Primignani M, Chantarangkul V et al. An Imbalance of pro- vs Group. The International Normalized Ratio (INR) in the MELD score:
anti-coagulation factors in plasma from patients with cirrhosis. Gastro- problems and solutions. Am J Transplant 2010;10:1349–53.
enterology 2009;137:2105–11. 58. Bellest L, Eschwege V, Poupon R et al. A modified international normalized
43. Gatt A, Riddell A, Calvaruso V et al. Enhanced thrombin generation in patients ratio as an effective way of prothrombin time standardization in hepatology.
with cirrhosis-induced coagulopathy. J Thromb Haemost 2010;8:1994–2000. Hepatology 2007;46:528–34.
44. Lisman T, Bakhtiari K, Pereboom IT et al. Normal to increased thrombin 59. Marlar RA. Determining the model for end-stage liver disease with
generation in patients undergoing liver transplantation despite prolonged better accuracy: neutralizing the international normalized ratio pitfalls.
conventional coagulation tests. J Hepatol 2010;52:355–61. Hepatology 2007;46:295–6.
45. Caldwell SH, Hoffman M, Lisman T et al. Coagulation disorders and hemo- 60. Shah NL, Intagliata NM, Northup PG et al. Procoagulant therapeutics
stasis in liver disease: pathophysiology and critical assessment of current in liver disease: a critique and clinical rationale. Nat Rev Gastroenterol
management. Hepatology 2006;44:1039–46. Hepatol 2014;11:675–82.
46. Tripodi A, Anstee QM, Sogaard KK et al. Hypercoagulability in cirrhosis: 61. Gunawan B, Runyon B. The efficacy and safety of epsilon-aminocaproic
causes and consequences. J Thromb Haemost 2011;9:1713–23. acid treatment in patients with cirrhosis and hyperfibrinolysis. Aliment
47. Northup PG, McMahon MM, Ruhl AP et al. Coagulopathy does not fully Pharmacol Ther 2006;23:115–20.
protect hospitalized cirrhosis patients from peripheral venous thrombo- 62. Ferro D, Celestini A, Violi F. Hyperfibrinolysis in liver disease. Clin Liver
embolism. Am J Gastroenterol 2006;101:1524–8. Dis 2009;13:21–31.
48. Søgaard KK, Horváth-Puhó E, Grønbaek H et al. Risk of venous throm- 63. Lisman T, Leebeek FW, Mosnier LO et al. Thrombin-activatable fibrinolysis
boembolism in patients with liver disease: a nationwide population-based inhibitor deficiency in cirrhosis is not associated with increased plasma
case-control study. Am J Gastroenterol 2009;104:96–101. fibrinolysis. Gastroenterology 2001;121:131–9.
49. Englesbe MJ, Schaubel DE, Cai S et al. Portal vein thrombosis and liver 64. Colucci M, Binetti BM, Branca MG et al. Deficiency of thrombin activat-
transplant survival benefit. Liver Transpl 2010;16:999–1005. able fibrinolysis inhibitor in cirrhosis is associated with increased plasma
50. Villa E, Cammà C, Marietta M et al. Enoxaparin prevents portal vein fibrinolysis. Hepatology 2003;38:230–7.
thrombosis and liver decompensation in patients with advanced cirrhosis. 65. Rijken DC, Kock EL, Guimarães AH et al. Evidence for an enhanced
Gastroenterology 2012;143:1253–60. fibrinolytic capacity in cirrhosis as measured with two different global
51. Wanless IR, Wong F, Blendis LM et al. Hepatic and portal vein thrombosis fibrinolysis tests. J Thromb Haemost 2012;10:2116–22.
in cirrhosis: possible role in development of parenchymal extinction and 66. Northup PG, Caldwell SH. Coagulation in liver disease: a guide for the
portal hypertension. Hepatology 1995;21:1238–47. clinician. Clin Gastroenterol Hepatol 2013;11:1064–74.
52. Tripodi A, Fracanzani AL, Primignani M et al. Procoagulant imbalance in 67. De Pietri L, Bianchini M, Montalti R et al. Thrombelastography-guided
patients with nonalcoholic fatty liver disease. J Hepatol 2014;61:148–54. blood product use before invasive procedures in cirrhosis with severe
53. Tripodi A, Mannucci PM. Mechanisms of disease. The coagulopathy of coagulopathy: a randomized, controlled trial. Hepatology 2016;63:
chronic liver disease. New Engl J Med 2011;365:147–56. 566–73.
54. Trotter JF, Brimhall B, Arjal R. Specific laboratory methodologies achieve 68. Intagliata NM, Caldwell SH, Porte RJ et al. Prediction of bleeding in
higher model for endstage liver disease (MELD) scores for patients listed cirrhosis patients: is the forecast any clearer? Hepatology 2015;64:
for liver transplantation. Liver Transpl 2004;10:995–1000. 989–90.

© 2017 by the American College of Gastroenterology The American Journal of GASTROENTEROLOGY