Cirrhosis of the liver frequently results in substantial changes in the hemostatic system. A decreased platelet count, impaired platelet function, a decreased thrombin generating capacity, defective fibrin formation because of dysfibrinogenemia, and defects in the fibrinolytic system may all be encountered in these patients (reviewed in ). The net effect of these hemostatic changes is thought to be a bleeding tendency, which may be particularly manifested during invasive procedures. Platelet function may be impaired in cirrhosis because of multiple direct and indirect mechanisms. A low hematocrit, a low platelet count, and increased production of the platelet inhibitors nitric oxide and prostacyclin are presumably responsible for a decreased platelet response in vivo. Functional defects of the platelet itself have been described but remain poorly defined (reviewed in ). A recent number of publications revisited the hemostatic abnormalities in patients with liver disease and indicated that, although many of the hemostatic changes involve defects in pro-hemostatic systems, compensatory mechanisms are also found. These compensatory mechanisms include substantially elevated levels of von Willebrand factor (VWF), which may compensate for defects in platelet number and function, decreased levels of anticoagulants, and decreased levels of antifibrinolytic proteins [2–4]. Based on these publications, it is now believed that the hemostatic abnormalities in patients with liver disease result in a more or less (re)-balanced hemostasis, and that the hemostasis-related bleeding problems are less frequent and severe than previously thought [5–7]. These novel insights in the hemostatic abnormalities in patients with liver disease prompted us to revisit the platelet defects in these patients in an established model of platelet adhesion under flow conditions . The study was approved by the medical ethical committee of the Erasmus University Medical Center and written informed consent was obtained from all patients and controls. When citrated whole blood from three patients with Child C alcohol-induced cirrhosis, and from one patient with Child B cirrhosis induced by viral hepatitis (hematocrit 26.7% ± 1.8%, mean ± SD, mean platelet count 99 000 ± 42 000 μL−1) was perfused over collagen (at a shear rate of 800 s−1) or fibrinogen (at a shear rate of 300 s−1) for 5 min, we observed a substantial decrease in platelet deposition to both collagen (P = 0.02) and fibrinogen (P < 0.005) compared with platelet deposition using citrated whole blood from four healthy volunteers (hematocrit 42.1% ± 1.6%, platelet count 232 000 ± 32 000 μL−1), see Fig. 1. Hematocrit (P < 0.001) and platelet count (P = 0.002) were significantly lower in the patients as compared to the controls. To investigate whether the decreased platelet adhesion, in case of the patients with liver disease, was due to the reduced hematocrit and platelet count, or was also a consequence of functional platelet defects, we evaluated platelet adhesion to collagen and fibrinogen under experimental conditions in which platelet count and hematocrit of both patients and controls were standardized to a hematocrit of 40% and a platelet count of 200 000 μL−1 by washing procedures as described . Washed platelets from cirrhosis patients or controls were resuspended in a 4% human albumin solution (HAS) or autologous plasma and mixed with red cells to obtain reconstituted blood with a platelet count of 200 000 μL−1 and a hematocrit of 40% and perfused over collagen or fibrinogen. In selected experiments, the reconstituted HAS blood was preincubated with a mixture of clotting factors [rFVIIa (1.2 μg mL−1), factor (F) × (10 μg mL−1), and prothrombin (20 ng mL−1)] in the presence of calcium chloride (3 mm) for 5 min at 37 °C. Addition of these clotting factors (thrombin generating system) results in thrombin-dependent increase in platelet deposition . Perfusion experiments over collagen were performed for 5 min at a shear rate of 800 s−1, whereas perfusions over fibrinogen were performed for 5 min at a shear rate of 300 s−1. A total of 16 patients were used for this part of the study. Seven patients had alcohol-induced cirrhosis, four had primary sclerosing cholangitis, three had viral hepatitis, and two had cryptogenic cirrhosis. Nine patients with Child A cirrhosis, three patients with Child B cirrhosis, and four patients with Child C cirrhosis were studied. Eighteen healthy volunteers from our laboratory served as controls. Because a large amount of blood was required to perform these experiments, we did not have access to sufficient patient material to study all parameters in every patient. However, all conditions studied were performed in triplicate. The number between the brackets in Fig. 2 indicates the number of patients and controls studied for each condition. The surface coverage measured in collagen adhesion experiments is represented in Fig. 2A. Adhesion of platelets resuspended in HAS was similar between patients and controls (P = 0.52). After addition of a thrombin generating system containing recombinant FVIIa, FX, and prothrombin, platelet adhesion increased significantly in the control group (P = 0.008), whereas the increase in platelet adhesion in the cirrhosis group did not reach statistical significance (P = 0.08). However, the difference between platelet adhesion in the presence of the thrombin generating system was not statistically significant between patients and controls (P = 0.08). When platelets were resuspended in autologous plasma, no difference between adhesion of control and cirrhosis platelets was observed (P = 0.68). Figure 2C shows a typical example of platelet thrombi formed on collagen when patient or control platelets were resuspended in autologous plasma at a standardized platelet count and hematocrit. The surface coverage measured in fibrinogen adhesion experiments at standardized platelet count and hematocrit is represented in Fig. 2B. Adhesion of platelets resuspended in autologous plasma appeared higher in the cirrhosis group as compared to the control group, but this difference did not reach statistical significance (P = 0.08). Figure 2D shows a typical example of the morphological appearance of the platelet deposits. No differences between platelet adhesion to both collagen and fibrinogen were observed between patients with different etiology or Child status. This study shows that platelets from patients with liver cirrhosis are able to interact with fibrinogen and to form thrombi on collagen with similar efficacy as platelets from healthy controls, provided platelet count and hematocrit are adjusted to levels found in the healthy controls. This indicates that the intrinsic platelet defects in patients with cirrhosis that have been described in literature are not relevant under physiological flow conditions. As the platelet count in patients with liver disease is usually sufficiently high to support normal hemostasis, our results suggest that it is unlikely that platelet dysfunction is an important contributor to bleeding in these patients. Our in vitro model does, however, show substantially defective platelet adhesion in platelet whole blood, indicating that the combination of a low hematocrit and a low platelet count may contribute to defective platelet function in vivo. This may also suggest that correction of anemia may improve hemostatic capacity in cirrhotic patients, similar to hemostatic improvement of patients with uremia after red cell transfusion or treatment with erythropoietin [10,11]. We have used different experimental conditions to test multiple aspects of platelet functioning. Platelet adhesion to fibrinogen at the shear rate used in this study is fully dependent on the integrin αIIbβ3. As platelet deposition in reconstituted blood was similar (or perhaps even slightly elevated, see Fig. 2B) in patient blood as compared to control blood, we conclude that αIIbβ3 is functionally normal in cirrhosis. Platelet adhesion and aggregation to collagen in a reconstituted system either containing human albumin as a plasma substitute or autologous plasma was similar in patients and controls, indicating that the interplay between VWF, glycoprotein (GP) Ib, the collagen receptors α2β1 and GP VI, and αIIbβ3 leading to thrombus formation on collagen is not disturbed in patients with liver cirrhosis. Finally, we tested the effect of in situ thrombin generation on platelet deposition to collagen and observed a slightly impaired response in the patient group. This could indicate that platelets of cirrhosis patients have a reduced response to thrombin because of defective PAR-1 or -4 function or defective signaling downstream of the PARs. However, another possibility is that patient platelets are defective in procoagulant capacity as the thrombin response in our experimental set-up depends on thrombin generation via FVIIa on the collagen-adhered platelet . Our results are in contrast with the general concept of the presence of platelet function defects in patients with liver disease. However, most studies investigating platelet function in these patients used static systems or aggregometry studies (reviewed in ). Only one study examined platelet function using a model employing flow conditions in a patient population with similar characteristics to the population used in this study. Ordinas et al.  did find functional platelet defects independent of hematocrit and platelet count. The differences between their study and the study described in this article are the use of inverted rabbit aorta segments instead of purified human collagen and fibrinogen as adhesive surfaces, and the use of a recirculating perfusion system as compared to the single pass system used in the present study. The same group, however, has also examined platelet function in cirrhotic patients using the platelet-function analyzer, which is also a flow-based assay, and found that the elevated closure times of both the collagen/adenosine 5′-diphosphate and collagen/epinephrine cartridge could be completely normalized by adjustment of the hematocrit . These latter results thus are in agreement with the results presented in the present study.
Disclosure of Conflict of Interests
The authors state that they have no conflict of interest.