LAUSR

We thank Dr Rathinasamy and colleagues for their interest in our study and their comments. We have compared the hemostatic status in patients with acute-on-chronic liver failure (ACLF) with patients with sepsis without underlying liver disease to assess whether the hemostatic changes in ACLF are primarily an exaggeration of cirrhosis-associated hemostatic changes or whether hemostatic changes induced by critical illness (independent of underlying liver disease) also contribute.1 We found a remarkable overlap between the hemostatic status in ACLF and patients with sepsis without underlying liver disease, with the exception of a profound hyperfibrinogenemia accompanied by clots that are resistant to lysis in the latter group. We thus hypothesize that part of the hemostatic changes in patients with ACLF reflect the general hemostatic changes in patients with critical illness. Although we acknowledge that both ACLF and nonliver sepsis are very heterogenous diseases with associated heterogeneity in hemostatic status, we disagree with the statement that “since both groups have different hemostatic mechanisms comparing these two groups may not be appropriate” as it was our specific intention to catalog potential similarities between the groups. One reason for making this comparison is that many cases of ACLF have no demonstrable precipitant and covert/unconfirmed sepsis is often suspected; for this reason we feel this direct comparison is clearly clinically relevant and important to understanding pathophysiology. Dr Rathinasamy and colleagues further comment that patients with ACLF in our study appear in hemostatic balance, whereas other studies have described ACLF to be a hypocoagulable syndrome. The other studies referred to have used viscoelastic tests such as thromboelastography (TEG),2 rotational thromboelastometry (ROTEM),3 and the Sonoclot analyzer4 to estimate hemostatic capacity. Notwithstanding the advantages of whole blood hemostasis tests, there are particular limitations of viscoelastic tests in patients with liver disease that may not allow the conclusion that patients with ACLF are hypocoagulable.5 Specifically, viscoelastic tests are insensitive for plasma levels of von Willebrand factor (VWF) and protein C. However, patients with liver disease have profound increases in VWF plasma levels and profound decreases in protein C levels, which are most pronounced in the sickest patients, including those with ACLF. Because high VWF and low protein C plasma levels in patients with liver disease promote hemostasis and compensate for thrombocytopenia6 and decreased plasma levels of procoagulant proteins,7 respectively, viscoelastic tests underestimate hemostatic potential in patients with liver disease. In addition, the observation that viscoelastic test parameters become more hypocoagulable in septic patients and in sicker patients may be misleading. First, plasma levels of VWF and protein C become more abnormal (and more prohemostatic) in sicker patients,8 and again this is not taken into account by viscoelastic analyses. Second, although there are associations between more hypocoagulable viscoelastic test parameters and bleeding, this association may not be causal. Many bleeding complications in patients with ACLF are related to portal hypertension rather than to hemostatic failure, and the association between abnormal hemostatic parameters and risk of portal hypertension–related bleeding can also be explained by the fact that abnormal hemostasis tests reflect severity of disease. For example, platelet count, which is an important determinant of viscoelastic test results, is directly related to severity of portal hypertension. Dr Rathinasamy and colleagues ask for details on ACLF grading and occurrence of sepsis in the ACLF group. About half of the patients had ACLF grade 1, whereas a quarter (each) had grade 2 and 3. Almost half of the patients had infection on admission. There were no clear differences in hemostatic parameters according to ACLF grade or sepsis status, but the subgroups were too small for meaningful analyses. We have previously investigated and reported on differences between acutely decompensated cirrhosis and ACLF as also requested, and found notable differences which underline that ACLF is a unique clinical syndrome.8,9 Again, because of the specific hemostatic changes associated with ACLF, we performed a study to assess whether the hemostatic changes in ACLF are an exaggeration of those found in (decompensated) cirrhosis, or whether hemostatic changes related to critical illness contribute. We concur with Dr Rathinasamy and colleagues that our study is limited by the small sample size and inevitable heterogeneity in the cohorts, but do feel that our data indicate that hemostatic balance is largely maintained both in patients with ACLF and non-liver sepsis. This finding has implications for clinical management. Both in ACLF and non-liver sepsis a restrictive use of procoagulants is indicated, even in patients with profound abnormalities in routine diagnostic tests of hemostasis including viscoelastic tests. Conversely, anticoagulant therapy might be beneficial, even in patients that appear