If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Reprint requests Address requests for reprints to: Alberto Queiroz Farias, MD, Hospital das Clínicas da Faculdade de Medicina da Universidade de São Paulo, Av. Dr. Eneas de Carvalho Aguiar, 255, sala 9159, São Paulo, SP, Brazil, Postal code 05403-000. fax: (55) 11-30621109
There is controversy over whether coagulation status predicts bleeding caused by ulceration after esophageal varices band ligation (EVL).
EVL was performed for primary (n = 45) or secondary (n = 105) prophylaxis in 150 patients with cirrhosis (Child A, n = 74, 49%; Child B, n = 42, 28%; Child C, n = 34, 23%). International normalized ratio (INR) and platelet counts were assessed in all. In 92 patients, levels of factor V, fibrinogen, D-dimer, protein C and protein S, von Willebrand factor, and thromboelastography (TEG) were assessed. Platelet count <50 × 103/mm3 and INR >1.5 were considered high-risk cutoff for bleeding. Conversely, platelet count ≥50 × 103/mm3 with INR ≤1.5 were safe cutoffs.
Overall, 11 patients (7.3%) had post-EVL ulcer bleeding. Bleeding occurred in 5 patients with Child A/B (4.3%) and 6 patients with Child C (17%) (P = .0174 for Child A/B versus Child C). Eight patients with bleeding were among the 110 below the cutoff for INR and platelet count, whereas only 3 of the patients with bleeding were among the 40 patients with purported high-risk values (P = 1.0). Among the 92 patients with expanded coagulation tests, bleeding occurred in 5. There was no difference in any of the coagulation parameters, including overall TEG patterns, between patients who did and did not bleed.
Post-EVL ulcer bleeding was associated with Child C status but not with conventional or expanded coagulation indices in cirrhotic patients without renal failure or infection undergoing elective EVL. These results call into question the common use of prophylactic procoagulants in the elective setting.
See related article, Kumar A et al, on page 892 in Gastroenterology.
The overall incidence of variceal hemorrhage in cirrhotic patients without prophylaxis ranges from 25%–35% during a period of 2 years, whereas rebleeding rates are around 40%–70% in early series and 10% in a more recent study.
According to current guidelines, esophageal varices band ligation (EVL) is considered first-line therapy to control acute bleeding, to prevent recurrence, and to prevent first bleeding episode when beta-blockers are contraindicated or not tolerated.
Portal hypertension and variceal bleeding–unresolved issues: summary of an American Association for the study of liver disease and European Association for the study of the liver single-topic conference.
Because no comprehensive study has been performed, it is disputed whether poor coagulation as measured conventionally is associated with increased frequency of post-EVL ulcer bleeding. On the other hand, current concepts on coagulation in liver cirrhosis have challenged the role of impaired coagulation as a risk factor for bleeding after invasive procedures and suggested that more elaborate tests might better capture bleeding risk.
Liver disease down-regulates the procoagulant system (clotting factors and platelets) and the anticoagulant system (natural anticoagulant protein C and protein S). The parallel decline of both mechanisms often leads to rebalanced hemostasis.
In fact, it has been reported that in spite of abnormal coagulation tests, thrombin generation might be normal in cirrhosis, and elevated von Willebrand factor levels support platelet adhesion even in the presence of reduced functional capacity.
In addition, it was shown that conventional coagulation tests such as prothrombin time and international normalized ratio (INR) assess no more than 5% of thrombin generation, measure only the activity of procoagulants, and fail to capture changes in anticoagulants. Therefore, it is proposed as a poor predictor of bleeding in cirrhosis. Those in vitro lines of evidence suggest that compensatory mechanisms in cirrhosis serve as a counter force to preserve functional coagulation capability.
Thus, the bleeding risk after invasive procedures might not be predicted by conventional coagulation tests that fail to assess the complex interactions between the systems involved in the rebalancing of the coagulopathy in cirrhosis. Our aim was to assess the predictive value of conventional and expanded coagulation parameters on postbanding bleeding. The primary end point was significant postbanding ulcer bleeding within 30 days, defined by Baveno IV criteria.
Patients and Methods
From April 2006 to July 2008, all cirrhotic patients (n = 161) referred to EVL in the Department of Gastroenterology of the University of Sao Paulo School of Medicine were enrolled to participate in the current study. One hundred fifty patients met inclusion criteria as shown in Figure 1. Inclusion criteria were (1) liver cirrhosis diagnosed according to clinical, laboratory, and radiologic data and/or biopsy-proven; (2) age older than 18 years old. Exclusion criteria were (1) severe pulmonary and cardiac disease; (2) advanced hepatocellular carcinoma according to Milan criteria,
Patients were divided into 2 groups according to conventional cutoff values of platelet count (<50 × 103/mm3, n = 18; ≥50 × 103/mm3, n = 132) and INR (>1.5, n = 28; ≤1.5, n = 122). Among the 28 patients with increased INR, 13 had INR >1.7. Platelet count ≥50 × 103/mm3 and/or INR ≤1.5 were considered safe cutoffs. Conversely, platelet count <50 × 103/mm3 and/or INR >1.5 were considered high-risk cutoffs. Those values represent commonly used cutoffs for relative contraindication of invasive procedures.
Esophageal Varices Band Ligation
Endoscopy was performed by using standard technique, after an overnight fast, with conscious sedation (midazolam and fentanyl), by using the 6 shooter SAEED multi-band ligator (Wilson Cook Co, Winston-Salem, NC). Up to 6 bands were placed per session, which were repeated every 2 weeks until varices eradication. To achieve eradication, 2.5 ± 1.2 sessions (range, 1–7) were required. All patients were prescribed omeprazole 20 mg orally twice a day until 30 days of the last EVL session. All EVL sessions were performed on elective basis (no case of active bleeding). In our standard practice, patients who had variceal bleeding are prescribed prophylactic antibiotics until 7 days after the bleeding.
In 105 (70%) patients, the indication of EVL was prevention of recurrent bleeding, 90 of whom were on propranolol therapy, whereas in 45 (30%), the indication was primary prophylaxis.
Conventional and Expanded Coagulation Tests
Prothrombin time (PT), activated partial thromboplastin time (APTT), and platelet count were assessed in all patients. Prothrombin time was determined according to World Health Organization guidelines
by using the kit PT-Fibrinogen HS Plus (HemosIL; Instrumentation Laboratory, Lexington, MA), according to the recommendations of the manufacturer. Results were expressed as the INR. APTT was measured in citrated plasma by using an automated method, with the APTT-SP kit–liquid (HemosIL), and results were expressed as ratios of tests to reference plasma. An automated photo-optical coagulometer (Coag-A mate MTX; Organon Teknika, Oss, Netherlands) was used for measuring PT, APTT, and fibrinogen levels. Freshly collected blood samples drawn from a peripheral vein were anticoagulated with ethylenediaminetetraacetic acid and tested by an automated hematology analyzer for platelet count.
Ninety-two consecutive patients underwent expanded coagulation tests with factor V, fibrinogen, D-dimer, protein C and protein S, and von Willebrand factor. Those patients also had whole blood samples obtained immediately before EVL for thromboelastography (TEG) (Haemoscope TEG 5000 Analyzer, Niles, IL) by using citrated and recalcified plasma and sent for analysis within 1 hour. For TEG interpretation, we assessed the pattern and individual parameters R (time to initial fibrin formation), K (rapidity of clot formation up to 20 mm amplitude), MA (maximal amplitude, ie, clot strength, representing the dynamics of fibrin and platelet bonding), alpha angle (clot strengthening, rapidity of fibrin buildup), LY30 (measure of the rate of amplitude reduction 30 minutes after maximal amplitude), and G (exponential conversion of maximal amplitude to dyn/cm2). TEG patterns were classified as normocoagulability, hypocoagulability, and hypercoagulability on the basis of a composite score called the clotting index.
Protein C was measured as functional activity by using an automated chromogenic assay (Berichrom Protein C; Dade Behring, Newark, NJ). Protein S was measured as functional activity by using the commercially available kits (Protein S Ac; Dade Behring). Factor V activity was measured by using a standard technique of human plasma immunodepleted of factor V (HemosIL) based on the prothrombin time assay. Semiquantitative determination of D-dimer was carried out by latex agglutination assay (Amax Accuclot; Trinity Biothec, Wicklow, Ireland). Quantitative determination of fibrinogen on the basis of the Clauss method in human citrated plasma was performed with the kit Fibrinogen-C (HemosIL). Von Willebrand factor antigenic activity was assessed by in-house enzyme-linked immunoadsorbent assay, with results expressed as ratios of tests to reference plasma. In brief, rabbit anti-von Willebrand factor antibody (A0082; Dako, Glostrup, Denmark) coated microplates (Nunc, Roskilde, Denmark) were incubated with patient serum samples. The secondary antibody was the peroxidase conjugated rabbit anti-human von Willebrand factor (P226; Sigma Inc, St Louis, MO). The reaction was revealed with ortho phenylenediamine and read with the wavelength of 492 nm.
Follow-Up and End Points
There were 3 dropouts during follow-up, as shown in Figure 1. Follow-up endoscopy was performed 3 months after varices eradication in all others. Post-EVL ulcer bleeding was defined as any clinically significant episode of hematemesis, melena, or both on the basis of Baveno IV definitions
and confirmed by endoscopy. Other causes of bleeding were ruled out. A history of repetitive epistaxis, mucocutaneous bleeding, or hematomas was obtained by reviewing the clinical charts.
The study was carried out according to the standards of Helsinki declaration. After approval of protocol by the local Ethic Committee, an informed consent was obtained from all patients.
Student t test and the Kruskal–Wallis test were used to compare means of continuous variables. Variables with statistical significance (P < .05) were selected for inclusion in a logistic regression model. Chi-square test and Fisher exact test were used to compare dichotomous variables when appropriate. Spearman rank correlation was used to assess platelet count and G parameter. All calculations were performed with the software Prism 4 and Instat 3 (GraphPad Inc, La Jolla, CA) and the R software (2.8.1; http://www.r-project.org). Results were expressed as the mean ± standard deviation.
The overall incidence of severe post-EVL ulcer bleeding was 7.3% (n = 11). The interval between the last EVL session and the bleeding was 9.4 ± 4.5 days. There was no difference in the number of bands placed in bleeders and in nonbleeders (4.9 ± 1.1 versus 4.4 ± 1.0, P = .1364). Large size varices were found in 6 (55%) and 85 (61%) patients with and without postbanding ulcer bleeding, respectively (P = .7525), whereas red color signs were found in 10 (91%) and 129 (93%) patients, respectively (P = .5803). Active alcohol intake was reported in 1 patient of each group (P = .1417).
Bleeding was observed in Child A, n = 4 (5%); Child B, n = 1 (2%); and Child C, n = 6 (17%) patients (P = .0174 for Child A/B versus Child C status; odds ratio [OR], 4.7; 95% confidence interval [CI], 1.369–16.666). The OR for postbanding bleeding in patients with a Child–Pugh score 5 was 3.37% (95% CI, 1.13%–9.59%). Every point in Child–Pugh score increased the risk of bleeding 1.3723 times (95% CI, 1.0379–1.8145).
Post-EVL ulcer bleeding occurred independently of the coagulation test results, as shown in Table 2. Eight (7.2%) bleeders were among the 110 patients with safe cutoff of INR and platelet count, whereas 3 (7.5%) bleeders were included among the 40 patients who had purported high-risk values for INR and platelet count (P = 1.0). One patient had bleeding out of the 18 patients with platelet count <50 × 103/mm3, whereas 10 patients had bleeding among the 132 patients with platelet count ≥50 × 103 /mm3 (P = 1.0; OR, 0.7; 95% CI, 0.08–5.96). Three bleeding patients were seen among the 28 participants with INR >1.5, and the remaining 8 cases were among the 122 patients with INR ≤1.5 (P = .4310; OR, 1.7; 95% CI, 0.42–6.90). Four bleeding patients were among the 28 participants with APTT ≥1.2, and the remaining 7 cases were among the 122 patients with APTT <1.2 (P = .1248; OR, 2.7; 95% CI, 0.74–10.1). The association of Child C status with an elevated INR (OR, 7.9) or APTT (OR, 12.8) increased the risk of post-EVL bleeding. Child C class associated with low platelet threshold had an OR of 3.3.
Table 2Post-EVL Ulcer Bleeding According to Child–Pugh Status, Platelet Count, Levels of INR and APTT
Transfusion requirements were similar between patients with safe and risk value of INR and platelet count (5.0 ± 1.0 versus 5.5 ± 0.7 red cell packs; P = .5908).
Among the 92 patients with expanded coagulation tests, bleeding occurred in 5. There were no differences in levels of factor V, fibrinogen, protein C, protein S, and von Willebrand factor between bleeders and nonbleeders, as shown in Table 3.
Table 3Post-EVL Ulcer Bleeding According to Results of Expanded Coagulation Tests
The pattern of TEG was normocoagulability (n = 16, 1 bleeder), hypocoagulability (n = 55, 3 bleeders), and hypercoagulability (n = 21, 1 bleeder). There was no significant difference in the frequency of bleeding according to the pattern of TEG (P = 1.0 for hypocoagulability versus normocoagulability and hypercoagulability). Hyperfibrinolysis was not detected by TEG in this series. No single TEG parameter was predictive of bleeding risk, as shown in Table 4.
Table 4Post-EVL Ulcer Bleeding According to the Parameters of TEG
INR >1.5 was found in 2 (12%) of the 16 patients with normocoagulability, 18 (32%) of the 55 with hypocoagulability, and 4 (19%) among the 21 with hypercoagulability (P = .0935 for hypocoagulability versus normocoagulability and hypercoagulability), indicating that in spite of the increased INR, nearly one fourth of patients had normal or hypercoagulable TEG. On the other hand, platelet count correlated with the G parameter (Spearman r = 0.643; P < .0001), and values <50 × 103 were found in 15 (27%) individuals with a hypocoagulable TEG pattern, but not in patients with normal or hypercoagulable TEG.
Only one (9%) of the 11 patients with postbanding ulcer bleeding and 5 (3.5%) of the 139 without bleeding were found to have previous clinical history of mucocutaneous bleeding or spontaneous hematomas (P = .3719).
In patients who underwent primary and secondary prophylaxis, respectively, no significant difference was found in Child–Pugh C frequency (n = 11, 24% versus n = 23, 22%; P = .8318), INR >1.5 (n = 9, 20% versus n = 19, 18%; P = 1.0), platelet count <50 × 103 (n = 4, 9% versus n = 14, 13%; P = .5871), frequency of hypocoagulable TEG pattern (n = 14, 31% versus n = 41, 40%; P = .2108), and postbanding ulcer bleeding (n = 3, 7% versus n = 8, 8%, P = 1.0).
The concept that coagulation disorders in cirrhosis increase the risk of bleeding has been accepted in hepatology for many decades, and coagulation tests have been used to identify patients at high risk, to guide hemostatic therapy, or even to contraindicate invasive procedures. Nevertheless, conventional coagulation tests do not take into account either the rebalanced coagulation in cirrhosis that can result in normal thrombin generation or the influence of common factors that modify the equilibrium between the antagonic procoagulant and anticoagulant systems, such as the release of heparinoid factors during systemic infections.
From this perspective, conventional coagulation tests would be of limited value in predicting outcome. Conversely, pharmacologic therapy of coagulation disorders, before invasive procedures in cirrhosis, has not been proved to benefit patients. In this regard, a recent meta-analysis
found no evidence that the use of human recombinant activated factor VII reduces the risk of death in patients with liver disease and upper gastrointestinal bleeding. Furthermore, the Baveno IV consensus on Portal Hypertension
has stated that currently available data are insufficient for recommendations regarding the management of coagulopathy and thrombocytopenia during acute variceal bleeding.
In the current study, we assessed the relationship of post-EVL ulcer bleeding and coagulation tests in cirrhotic patients without concurrent conditions that interfere in blood coagulation. The overall frequency of postbanding ulcer bleeding was similar to those reported elsewhere,
but we failed to demonstrate a relationship between INR, platelet count and APTT, and bleeding. Even when we grouped patients into safe and high-risk groups, we found no significant difference between them in the frequency of post-EVL ulcer bleeding.
Although we are aware that normal thrombin generation has been reported in vitro above the platelet threshold of 56 × 103/mm3, patients were divided into groups according to the common cutoff of 50 × 103/mm3, usually used in clinical practice.
Thus, it is likely that platelet function in vivo is maintained even in the presence of lower platelet count.
Expanded coagulation testing with fibrinogen, factor V, protein C, protein S, D-dimer, and von Willebrand factor also could not differentiate bleeders and nonbleeders. Obtained values reflected the severity of underlying cirrhosis, with the rebalanced coagulation characterized by a reduction of factor V and fibrinogen and natural anticoagulants protein C and protein S. It is noteworthy that D-dimer levels were increased but did not correlate with bleeding or with hyperfibrinolysis. Recent data suggest that high D-dimer is a marker of severity of cirrhosis with ascites.
In fact, 22% of patients in this series had advanced disease classified as Child C. High plasma levels of von Willebrand factor are a marker of circulatory dysfunction in cirrhosis and associated with restoration of platelet adhesion, counterbalancing the reduction in platelet count.
This might explain the lack of difference in bleeding rates, because higher levels were observed in both groups. The data do suggest that predictive patterns might emerge in subgroups such as the Child C patient, but the number was too small in this study to draw firm conclusions.
TEG is a method that enables global assessment of plasmatic and cellular components of coagulation in freshly or citrated whole blood, including the procoagulant and anticoagulant factors, platelet function, and hyperfibrinolysis. Although the value of TEG is established only in the setting of liver transplantation and cardiopulmonary surgery, this method has gained applicability for investigating the coagulation status in conditions in which standard tests have failed to unravel.
In this series post-EVL bleeding occurred regardless of TEG pattern, and no isolated TEG parameter could predict bleeding. We did not find hyperfibrinolysis, which is estimated to occur in up to 31% of decompensated cirrhosis patients in the current series
and which has been proposed as a mechanism of bleeding after invasive procedures. No relationship between elevated INR and TEG pattern was observed in the current study, because abnormal INR was seen in patients with hypocoagulable TEG pattern, normal and hypercoagulable patterns, without significant difference. Low platelet threshold correlated with hypocoagulable TEG pattern and G parameter, but neither parameter could predict postbanding ulcer bleeding.
On the other hand, decreased liver function, assessed by the Child–Pugh classification, was the only factor associated with post-EVL bleeding. Child C status was found to be 3 times more frequent in bleeders than in nonbleeders and increased the bleeding risk almost 5 times. Those findings are in accordance with Yang et al,
who found bacterial infection and end-stage liver cirrhosis (Child C) to be independent risk factors for early bleeding post-EVL. In addition, different authors have shown that the hepatic venous gradient pressure is higher in Child C patients when compared with Child A and Child B, and this difference could explain the higher frequency of post-EVL ulcer bleeding in that subset of patients.
The North Italian Endoscopic Club for the Study and Treatment of Esophageal Varices Prediction of the first variceal hemorrhage in patients with cirrhosis of the liver and esophageal varices: a prospective multicenter study.
In the current article, we investigated whether medical history of bleeding diathesis could predict the outcome after EVL, because primary hemostatic defects tend to produce bleeding in high shear-pressure systems with mucocutaneous bleeds, whereas secondary disturbances are associated with bleeding in low pressure areas, causing hematomas.
However, we could not demonstrate the relationship between previous clinical history and the increased risk of postbanding ulcer bleeding, indicating that neither clinical nor laboratory work-up used to assess procoagulant and anticoagulant factors and hyperfibrinolysis was able to characterize individuals at risk of severe ulcer bleeding.
We are aware of the limitations of the current study. Patients with concurrent medical conditions that interfere with coagulation, in particular renal failure, spontaneous bacterial infection, and anticoagulant use, were excluded. Thus, our conclusions might not be applicable to all patients who undergo EVL in clinical practice. We could not assess additional tests or emerging technologies, such as platelet aggregation studies, flow cytometry procedures, and thrombin generation potential, in predicting post-EVL ulcer bleeding. Nevertheless, those tests have intrinsic limitations and standardization issues. The use of plasmas from patients with liver diseases instead of routine use of plasmas from patients on oral anticoagulation in the calibration for thromboplastins was proposed for the development of a specific INR for liver disease (INR-liver) to achieve a better standardization of prothrombin time.
We did not investigate the applicability of the INR-liver for predicting procedure-related hemorrhage. Nevertheless, the INR-liver has been shown to significantly influence Model for End-Stage Liver Disease calculation,31 but, to our knowledge, it has never been used to assess bleeding risk. Finally, the sample size and number of outcomes were small to rule out completely the relevance of impaired coagulation on post-EVL bleeding. Considering the composite parameter INR and platelet count, the assessment of the bleeding risk, with an estimated power set at 0.8, would have required the inclusion of 16,608 cases and 199,296 controls. Those figures seem unrealistic for a single center and warrant consideration of future multicenter studies with very large sample sizes.
In summary, bleeding from post-EVL ulceration in the elective setting was not associated with demonstrable abnormalities of coagulation as assessed by conventional or expanded testing in this series. Given the limitations inherent to these tests, it is difficult to endorse specific cutoffs as a guide for performance or administration of prophylactic agents such as plasma, which carries its own inherent risk, although postprocedure bleeding remains a significant source of morbidity. Further work is necessary to explore alternative means of measuring hemostasis, the effects of procoagulants and the role of more severe portal hypertension, and the interaction between these 2 parameters as demonstrated by the work of Bosch et al,
Portal hypertension and variceal bleeding–unresolved issues: summary of an American Association for the study of liver disease and European Association for the study of the liver single-topic conference.
Endoscopic variceal ligation (EVL) and propranolol are standard secondary prophylaxis therapies for variceal bleeding. Addition of isosorbide mononitrate (ISMN) to propranolol improves its hemodynamic efficacy; we investigated whether a combination of EVL and propranolol/ISMN was more effective than EVL alone for secondary prophylaxis.