Bleeding Complications Associated with the Molecular Adsorbent Recirculating System: a retrospective observational study

Background: The molecular adsorbent recirculating system (MARS) is an artificial liver support system that supports excretory liver function in patients with liver failure and is used as bridge therapy for patients waiting for liver transplantation. However, MARS may increase the tendency for bleeding. The objective of this study was to determine how MARS affects coagulopathy and identify specific factors associated with bleeding complications. Methods: We retrospectively analyzed data from 15 patients undergoing a total of 36 MARS sessions. Complete blood count, coagulation profiles, and blood chemistry values were compared before and after MARS. To identify pre-MARS factors associated with increased bleeding after MARS, we divided patients into bleeder and non-bleeder groups and compared their pre-MARS laboratory values. Results: MARS significantly reduced bilirubin and creatinine levels. MARS also increased prothrombin time and activated partial thromboplastin time and reduced fibrinogen, thus negatively impacting coagulation. Seven patients had bleeding complications and were classified into the bleeder group. Pre-MARS hemoglobin was significantly lower in the bleeder group (8.3 mg/dl) than in the non-bleeder group (10.0 mg/dl, P=0.014). When comparing the upper and lower 25% of MARS sessions based on the hemoglobin reduction rate, hemoglobin reduction was significantly greater in MARS sessions involving patients with low pre-MARS hemoglobin and factor V (P=0.008 and P=0.032, respectively). Conclusions : MARS appears to alter coagulation-related factors and increase the risk of bleeding complications. However, individual differences among patients were large, and various factors, such as low hemoglobin and factor V levels, appear to be involved.


BACKGROUND
The treatment of acute liver failure encompasses symptomatic supportive care as a medical treatment and liver transplantation as a surgical treatment. The need for intensive critical care is based on the severity of symptoms, such as hemodynamic instability, encephalopathy, bleeding, and hepatorenal syndrome [1]. Liver transplantation should be considered upon progression to decompensated liver failure [2]. However, as liver failure progresses rapidly and exhibits a varying course, it is difficult to make early decisions regarding liver transplantation in practice. In addition, finding living donors for liver transplantation is difficult, and obtaining appropriate cadaveric donors takes a considerable amount of time.
The molecular adsorbent recirculating system (MARS) is an artificial liver support system that supports excretory liver function in patients with liver failure. It is widely used as a bridge therapy for patients waiting for liver transplantation [3,4]. MARS improves hepatic encephalopathy, cerebral blood flow, renal function, and systemic hemodynamics [5].
Although it may be a good option for patients with end-stage liver failure [6], many clinicians are reluctant to apply MARS due to its high price, various side effects, and lack of evidence that it reduces mortality and improves survival rate [7][8][9]. Adverse effects of MARS include infection, hypotension, severe coagulopathy, bleeding, respiratory failure, cardiac failure, acute pancreatitis, severe thrombocytopenia, and seizure. Of these, bleeding complications are closely linked to mortality among liver failure patients [10], and even if donors are obtained, bleeding complications can impose a major constraint on liver transplantation.
Since MARS has been employed in our hospital, bleeding complications have occurred in many patients during or after MARS, even though coagulation-preventing agents, such as heparin or nafamostat, are not administered. One patient died of bleeding complications, and many others could not proceed with additional MARS sessions. However, other patients showed no bleeding-related complications. Therefore, we sought to determine which patients are more likely to experience bleeding complications and any predictive factors. The objective of this study was to examine the association between MARS and bleeding complications, and to identify pre-MARS factors associated with bleeding complications.

Data collection
Data from all patients receiving MARS in the intensive care unit (ICU) at Jeonbuk National University Hospital between December 2016 and February 2020 were analyzed retrospectively, which included 15 patients undergoing a total of 36 MARS sessions. Laboratory data, hospitalization records, progress records, ICU record sheets, nursing records, transfusion records, and medication history were reviewed. The following data were collected: gender; age; predisposing factors; pre-existing liver disease or bleeding; computed tomography abdomen and biopsy findings; alpha-fetoprotein (AFP); Child-Turcotte-Pugh (CTP) score; Simplified University Hospital approved this study (approval number CUH 2020-03-064). As the study is a retrospective observational one, the need for patient consent was waived.

MARS procedure
MARS was considered when total bilirubin exceeded 30 mg/dl or increased rapidly and/or upon the appearance of hepatic encephalopathy or hepatorenal syndrome. In addition, due to its high cost, a comprehensive judgment was made considering factors such as the patient's age, possibility of transplantation, and the opinions of the patient or their family members.
Additional MARS sessions were considered based on total bilirubin levels and changes, complications and effects of previous MARS sessions, and further consent of the patient or their family members. The Baxter GAMBRO MARS ® kit was used, and heparin priming was performed. A Becton Dickinson™ arterial cannula was inserted into the radial or brachial artery, and an Arrow ® 's You-Bend™ two-lumen hemodialysis catheter and Arrow ® 's multilumen central venous catheter (7 French, 3-lumen) were inserted into the subclavian or femoral vein. No agent for preventing coagulation, such as heparin or nafamostat, was administered.

Data analysis
We first sought to determine the effect of MARS by comparing laboratory values before and after MARS. To control for temporal changes due to liver failure, the amount of change in values before and after MARS was compared with the amount of change in values between two time points of the same duration before MARS.
To identify pre-MARS factors associated with increased bleeding after MARS, we first divided patients into bleeder and non-bleeder groups, according to whether there was visually confirmed or strong suspicion of bleeding after MARS and compared their pre-MARS laboratory values. The bleeder group included patients with suddenly increased catheter oozing or ecchymosis, hemoglobin reduction by ≥2 g/dl, need for RBC transfusion with two or more packs, or a marked increase in bleeding after MARS. The non-bleeder group consisted of patients not included in the bleeder group (i.e., patients without catheter oozing or ecchymosis, hemoglobin reduction <2 g/dl after MARS, need for RBC transfusion with one or fewer packs, and no marked increase in bleeding after MARS). Second, we compared individual MARS sessions involving bleeder and non-bleeder patients (i.e., bleeder and non-bleeder sessions).
Third, we compared the upper and lower 25% of MARS sessions based on the amount of hemoglobin reduction after MARS, which were designated as bleeding and non-bleeding sessions, respectively.

Statistical analysis
Patient demographics and baseline values were compared using independent T tests or Chi square test, as appropriate. Kolmogorov-Smirnov and Shapiro-Wilk tests were used to determine whether data were normally distributed, after which groups were compared using independent T tests or Mann-Whitney nonparametric tests as appropriate. Changes over time in factors were tested using repeated measures analysis of variance (RM ANOVA). Statistical analysis was performed using IBM SPSS Statistics ver. 26 (SPSS, Chicago, IL, USA).

RESULTS
We examined data from 15 patients undergoing a total of 36 MARS sessions ( Table 1). The average age of patients was 49 years. Ten were men, and five were women. Four patients had acute liver failure (2 hepatitis A virus, 1 pregnancy-induced, 1 toxic hepatitis), and 11 patients had acute-on-chronic liver failure (8 alcoholic, 3 hepatitis B virus-related liver cirrhosis).
MARS was administered one to six times per patient. Seven of the 15 patients died, 7 were discharged and were on an outpatient follow-up, and 1 was transferred to another hospital.

Effects of MARS on laboratory values
The amount of change in laboratory values before and after MARS was compared across all patients ( Table 2). We found significant changes before and after MARS in total bilirubin, Fibrinogen increased by 4.7 mg/dl before MARS but decreased by 47.4 mg/dl after MARS (P=0.021). There were no significant differences in the amount of change before and after MARS in other chemistry profiles, complete blood count, or inflammatory markers. Figure 1 shows changes in coagulation factors measured before MARS, 2 hours after the start of MARS, and after MARS. Table 1 presents pre-MARS values for patients in the bleeder and non-bleeder groups. Pre-MARS hemoglobin was significantly lower in the bleeder group (8.2 mg/dl) than in the nonbleeder group (10.0 mg/dl; P=0.014). Apart from hemoglobin, however, there were no significant differences between patient groups. However, after classifying individual MARS sessions into bleeder and non-bleeder sessions, there were significant differences between groups in vasopressor use, pre-MARS hemoglobin, and pre-MARS factor V (Table 3). Bleeding complications were more common in MARS sessions involving vasopressor use (58.3%) than in those not involving vasopressor use (12.5%; P=0.045). In addition, pre-MARS hemoglobin and factor V were significantly lower in bleeder sessions than in non-bleeder sessions (8.3 vs. 9.5, P=0.001; 19.2 vs. 38.3, P=0.033, respectively). Table 4 shows a comparison of the upper and lower 25% of MARS sessions based on hemoglobin reduction rate. After MARS, hemoglobin decreased by 1.43 g/dl on average in the upper 25% of sessions (i.e., bleeding sessions) and increased by 0.29 g/dl in the lower 25% of sessions (i.e., non-bleeding sessions). Pre-MARS hemoglobin was significantly lower for bleeding sessions (8.5 mg/dl) than for non-bleeding sessions (10.2 mg/dl; P=0.008). Also, pre-MARS factor V was significantly lower for bleeding sessions (21.6%) than for non-bleeding sessions (45.5%; P=0.032). Other values did not significantly differ between groups.

DISCUSSION
Our results show that MARS acts as a liver support system by reducing total bilirubin, direct bilirubin, and creatinine. However, MARS may increase the risk of bleeding by increasing PT and aPTT and reducing fibrinogen. When comparing MARS sessions between bleeder and non-bleeder groups, parameters that differed significantly between groups were vasopressor use, pre-MARS hemoglobin, and pre-MARS factor V. Lower pre-MARS hemoglobin in the bleeding group suggests that bleeding, including micro-bleeding, may have been present prior to MARS, unless the patient had underlying anemia, chronic kidney disease, or hematological disease. There were no differences in pre-MARS parameters between bleeder and non-bleeder groups except for hemoglobin, and no patient had underlying chronic kidney disease. Thus, our results suggest that MARS increases bleeding risk when a tendency toward bleeding already exists. Pre-MARS platelet count, PT, and aPTT did not significantly differ between groups. As individual differences in bleeding tendency exist, laboratory values may not fully reflect a patient's bleeding tendency.
The liver is the site of synthesis of fibrinogen and factors II, V, VII, IX, X, XI, and XII. In patients with hepatic insufficiency, levels of these factors are low due to protein dysfunction and poor synthetic function. Defects in r-carboxyglutamic acid residues introduced by vitamin K-dependent carboxylase result in deterioration of the functions of factor ll, Vll, IX, and X and anticoagulant proteins C and S. By contrast, although factor V is mainly synthesized in the liver, it is a vitamin K-independent coagulation factor. As it has a relatively short plasma halflife of ~12 hours, it rapidly decreases in acute liver failure. Factor V is a major cofactor that converts prothrombin to thrombin and plays an important role in the coagulation pathway by regulating factor Ⅷ activity [11]. For this reason, many studies suggest that factor V levels are related to survival and can serve as a prognostic indicator [12][13][14].
Many studies suggest that MARS reduces bilirubin and creatinine and improves hepatic encephalopathy [4]. However, its side effects and survival rate are somewhat controversial.
According to a meta-analysis published in 2012 [5], MARS decreases total bilirubin and clinical symptoms of hepatic encephalopathy, but the incidence of bleeding complications was 24.1% compared with 10.2% among patients receiving standard medical support (P=0.007), and a consistent interpretation of mortality reduction across studies was difficult. In a metaanalysis published in 2019 [7], there was no difference in the survival of patients receiving standard medical support or MARS, but patients receiving high-intensity therapy with five or more MARS sessions showed better survival than those receiving low-intensity therapy with four or fewer sessions. Furthermore, a meta-analysis published in 2020 [15] reports that MARS reduces mortality (relative risk (RR), 0.84; 95% confidence interval (CI), 0.74-0.96; moderate certainty) among patients with acute or acute-on-chronic liver failure and is associated with bleeding-related side effects, such as hypotension (RR, 1.46; 95% CI, 0.98-2.20; low certainty), bleeding (RR, 1.21; 95% CI, 0.88-1.66; moderate certainty), and thrombocytopenia (RR, 1.62; 95% CI, 1.00-2.64; very low certainty). Bleeding complications of MARS can induce an unstable hemodynamic state, increase blood transfusion demand, and impede the procedure.
Bleeding complications can also promote disseminated intravascular coagulation in patients with liver failure, making medical staff hesitant to proceed with liver transplantation.
Many recent studies of MARS mention bleeding-associated side effects, such as thrombocytopenia, coagulopathy, hypofibrinogenemia, and anemia, but few studies focus on these outcomes. Among them, several studies using thromboelastography indicate that MARS induces platelet-mediated coagulopathy, both mechanically and immunologically [16,17].
MARS is an extracorporeal circulation system that potentially activates coagulation by putting blood in contact with artificial materials, resulting in the consumption of platelets and coagulation factors. We found that MARS reduced platelet count by 17500/µl as well as levels of coagulation factors. In particular, factor V was markedly reduced relative to other factors.
In addition, the large-bore venous cannulation required for MARS may be a factor that causes bleeding in liver failure patients with coagulopathy.
Unlike previous studies, we did not use heparin during MARS treatment and controlled the effects of blood transfusions such as RBCs, PC, and FFP. Central venous and arterial line catheters were inserted in all patients so that bleeding tendency could be evaluated under the same conditions. Above all, we compared coagulation factors and attempted to distinguish the effect of MARS from spontaneous coagulopathy caused by liver failure by comparing the amount of change rather than absolute laboratory values before and after MARS [18]. This study has several limitations. The number of patients was somewhat small, and the data collection period was long. Due to its retrospective design, the time that laboratory testing was performed differed slightly across patients. Moreover, there was an inevitable difference in timing because we had to select two time points between which blood transfusion was not performed. As we mainly classified patients into groups based on visible bleeding, researcher subjectivity cannot be completely excluded. The frequency of bleeding complications after MARS in this study was much higher than that in previous studies, even though heparin was not used. To explain this discrepancy, we reviewed the MARS procedure several times and contacted the MARS kit manufacturer to ensure that there were no changes in the kit or its recommended use instructions. We did not use thromboelastography and, therefore, could not evaluate platelet function, clot strength, or fibrinolysis.

CONCLUSION
MARS appears to alter coagulation-related factors, such as platelet count, PT, aPTT, fibrinogen, and coagulation factors, and increase the tendency toward bleeding complications in patients with liver failure. Therefore, the progression of coagulopathy should be considered when proceeding with MARS. However, the differences observed among patients were large, and various factors, such as vasopressor use and pre-MARS hemoglobin and factor V, appear to contribute to these differences. Further research on this subject is warranted given that the ability to predict bleeding complications, before their onset, would be useful to clinicians when making decisions regarding the use of MARS.

Research ethics and consent to participate
The Institutional Review Board of Jeonbuk National University Hospital approved this study (approval number CUH 2020-03-064). As the study is a retrospective observational one, the need for patient consent was waived.

Consent for publication
Not applicable.

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.