Albumin infusion reduces the risk of rebleeding and in-hospital mortality in cirrhotic patients admitted for acute gastrointestinal bleeding: retrospective analysis of a single institute


 BackgroundTo investigate the effect of albumin infusion in cirrhotic patients admitted for acute gastrointestinal bleeding.MethodsMedical records of cirrhotic patients who admitted due to acute gastrointestinal bleeding through January 2009 to December 2018 were systemically reviewed. Clinical data and the total amount of albumin and red blood cell used during hospitalization were recorded. For patients with rebleeding, the amount of albumin and red blood cell used before rebleeding was also documented. The primary outcome was the occurrence of rebleeding and the second outcome was in-hospital mortality. Univariate and multivariate logistic analysis were performed to identify risk factors associated with rebleeding and in-hospital mortality.ResultsA total of 2239 cirrhotic patients were included in the analysis. There were 245 episodes of in-patient rebleeding occurred, while 135 patients died. Overall, more red blood cells and albumin were prescribed to patients who suffered rebleeding. In terms of the amount before rebleeding, the red blood cell was higher in patients with rebleeding, but the albumin infusion were similar. In the multivariate model, the albumin infusion was an independent risk factor associated with rebleeding (adjusted OR for ≤40g, 0.385 [0.252-0.588], p<0.001; OR for >40g, 0.295 [0.169-0.514], p<0.001). The use of albumin more than 40g during hospitalization reduces the risk of in-patient mortality (adjusted OR for ≤40g, 0.730 [0.375-1.423], p=0.356; OR for >40g, 0.389 [0.180-0.838], p=0.016).ConclusionsAlbumin infusion could reduce risk of in-hospital rebleeding. Moreover, more than 40g albumin infusion decrease numbers of mortality in cirrhosis admitted for acute gastrointestinal bleeding.

3 may further deteriorate after GIB due to the direct loss of albumin in the gastrointestinal tract and short-term fast. As an established risk factor for sepsis and mortality in critical ill patients, hypoalbuminemia needs to be corrected by albumin treatment [1]. In patients with cirrhosis, the use of albumin prevents circulatory dysfunction after paracentesis, improves the outcomes in patients with refractory ascites, spontaneous bacterial peritonitis (SBP) and hepatorenal syndrome (HRS) [2,3].
However, no studies thus far have elucidated its role in the management of GIB in cirrhotic patients.
The antibiotic prophylaxis has been accepted by international consensus for its role in reducing rebleeding risk in cirrhotic patients [4]. The therapeutic effect of antibiotics in cirrhosis is attributed to the improvement of the inflammatory state caused by bacterial infection [5]. Meanwhile, emerging evidence supports the anti-inflammatory property of albumin, which contributes to reducing the risk of complications and mortality rate in cirrhotic patients [6,7]. On the contrary, albumin infusion, noted as a volume expander, may increase portal pressure and induce rebleeding due to deteriorated preexisting portal hypertension [8]. Therefore, our study aims to investigate whether the albumin infusion would affect the prognosis of cirrhotic patients admitted for GIB.

Study design
Medical records of cirrhotic patients who admitted to West China Hospital due to GIB through Jan 2009 to Dec 2018 were systemically reviewed. Data of patients' demographics, cause of cirrhosis, Child-Pugh classification, laboratory tests, days of hospital stay, sources of bleeding, initial rescue therapy with balloon tamponade, endoscopic and radiological interventions, concomitant of hepatic carcinoma (HCC) and portal vein thrombosis (PVT), occurrence of in-hospital rebleeding and deaths were collected based on the medical records.
The sources of bleeding were classified into upper or lower GI tract according to endoscopic findings.
The upper GI bleeding was further divided into variceal bleeding and non-variceal bleeding lesions as determined by esophagogastroduodenoscopy. Variceal bleeding was defined as active hemorrhage from varices, or the presence of a clot over varices, or varices as the only potential source of bleeding. The presence of HCC or PVT was confirmed by radiological imaging records or discharge 4 diagnosis. The baseline level of hemoglobin, bilirubin, albumin, creatine, and prothrombin time were obtained from the laboratory results at admission.

Clinical outcomes
The primary outcome was the in-hospital rebleeding event, defined as the following symptoms that occurred after initial hemostasis achieved: 1) newly occurrence of hematemesis or bloody nasogastric aspirate; 2) recurrence of hematochezia or melena. The secondary outcome was in-hospital mortality.
The death reasons were derived from the discharge diagnosis.

Infusion of albumin and red blood cell
The amount of albumin and red blood cell (RBC) infusion during hospitalization was collected from the hospital database. The total amount of albumin and RBC used during hospitalization were recorded as total-ALB and total-RBC. For patients with rebleeding occurred, the amount of albumin and RBC used before rebleeding was recorded as pre-ALB and pre-RBC. For patients without rebleeding, the amount of pre-ALB and pre-RBC was equivalent to the total-ALB and total-RBC.

Statistical analysis
Continuous variables were displayed as means ± standard deviation and compared using the Student's t test. Categorical variables were displayed as frequencies and percentages and compared by Chi-square or Fisher exact tests. Baseline characteristics and other associated covariates with outcomes were estimated with univariate and multivariate logistic regression models and reported as odds ratios (OR) with 95% confidence intervals (CI). Statistics are calculated using R version 3.2.0 (R Foundation for Statistical Computing, Vienna, Austria).

Patients' baseline characteristics and outcomes
A total of 2259 cirrhotic patients who admitted for acute GIB were selected, among whom 20 cases were excluded for incomplete data. Overall, 2239 records were included in this analysis. Overall, there were 245 (10.9%) patients for whom in-patient rebleeding occurred, with an average of 12.3±6.6 hospitalization days. As regards the in-hospital mortality, a total of 135 (6%) patients died during hospitalization and 110 of them experienced in-patient rebleeding. The reasons of deaths were as followed: hemorrhagic shock (n=95), hepatorenal syndrome (n=18), hepatic encephalopathy (n=7), liver failure (n=6), infection (n=4), cerebrovascular event (n=3), metastasis (n=1), acute myocardial infarction (n=1).

The association between albumin/RBC infusion for in-patient rebleeding
Univariate variables association with rebleeding were displayed in Table 1 To explore the association between the dose and risk of rebleeding, the amount of pre-ALB was further classified into two subgroups as ≤40g and > 40g. The use of pre-ALB was not linked to ( Table 3) Furthermore, in another multivariate model adjusted for all confounders (HCC, bleed from non-varices lesion, unidentified lesion, the initial application of balloon tamponade, level of bilirubin, albumin and prothrombin time, endoscopic treatments), the pre-ALB infusion was still associated with a lower risk of rebleeding, with a negative dose-effect relationship between albumin use and rebleeding risk  Figure 1B).

The association between albumin/RBC infusion and in-hospital mortality
In the univariate analysis, the in-hospital mortality was more probable in patients with compromised liver function (i.e. higher Child-Pugh classification, lower levels of albumin, higher values of bilirubin, creatine and prothrombin time), concomitant with HCC, bleed from varices or unidentified lesion, initial rescue therapy with tamponade and occurrence of rebleeding. The uses of endoscopic and radiological interventions, however, decreased mortality risk (  Figure 2B).

Discussion
Gastrointestinal bleeding (GIB) is a fatal complication of cirrhosis, especially for those with rupture of varices. The application of endoscopic and radiological interventions has greatly improved the prognosis of patients with acute GIB [9]. In addition to the advances of invasive treatments, the conservative treatments remain an integral part of the management of this medical emergency and is increasingly standardized. The therapeutic value of prophylactic antibiotics and restrict blood transfusion strategy has recently been proved by several clinical studies [5,8]. However, the role of albumin in the management of GIB in patients with decompensated cirrhosis has not yet been assessed. This study demonstrated that albumin infusion was associated with a lower risk of rebleeding and in-hospital mortality in cirrhotic patients.
Our study revealed a rebleeding rate of 10.8% and overall mortality of 6.0%, which were lower than those described in previous studies possibly due to a shorter follow-up period in our study [10,11].
Despite the risk factors that are already known for rebleeding, our study found that the unidentified lesion also increased the risk of rebleeding and in-hospital mortality in cirrhotic patients [5,11].
Current consensus recommended endoscopic intervention for non-variceal upper gastrointestinal bleeding within 24 hours since hemodynamic homeostasis, while the timing of endoscopy in cirrhosis suspected variceal bleeding seems not correlated to the outcomes [12]. Given the unclear urgency of endoscopic intervention and its requirement of technical expertise, endoscopy examinations for suspected variceal bleeding were only delivered after hemodynamic hemostasis in our institute.
However, based on our results, we advocated that appropriate endoscopic therapies for variceal lesion should be given within 5 days as most experts suggested. In consideration of the negative impact of unidentified lesion on the outcomes possible due to the rapid disease deterioration and passive surveillance to invasive treatments, we believe early identification of bleeding lesion is crucial for the management of acute GIB.
Our study aims to explore whether albumin infusion influences the outcomes of cirrhotic patients admitted for GIB. Theoretically, abundant albumin infusion in patients with GIB may cause rebleeding by increased portal pressure in a similar manner of liberal transfusion strategy [13]. Interestingly, we found that albumin infusion reduces the risk of in-hospital rebleeding. Particularly, this beneficial effect was predominately observed in patients with Child-Pugh C class, most of whom had high portal pressure.
In our study, the mean amount of pre-ALB was just 12.3 g, which is much lower than those recommended for ascites treatment (1-1.5 g/kg). Also, the dose-dependent effect of albumin in reducing rebleeding does not support the view that a higher dose leads to more bleeding. In fact, a recent study suggests that the portal pressure was not affected after the infusion of albumin (1.5 g/kg, every week) [6]. Hence, the concern that albumin infusion increases the risk of rebleeding due to deterioration of portal pressure seems unnecessary. On the other hand, albumin is used in cirrhosis for the treatment of SBP and was widely discussed in the sepsis treatment for its immunomodulatory and anti-inflammatory properties [14,15]. As bacterial infection was identified as a critical determinant of rebleeding in cirrhosis by increasing portal pressure through vasoactive substances [5], the effect of albumin on systemic inflammation provides a potential theoretical basis for its role in reducing bleeding. Moreover, albumin may also control the bleeding risk by promoting the transportation of drugs such as proton pump inhibitor and antibiotics, two essential agents in the management of acute GIB. Indeed, a recent clinical study found that the combination of albumin and antibiotics is superior to antibiotics alone in the control of inflammation [6]. Collectively, the mechanisms involved the benefit of albumin is attribute to its effect on systemic inflammation and pharmacokinetic process, although its other properties may also play a role [16,17].
The association between albumin infusion and in-hospital mortality was also evaluated. In our study, infection, HRS and liver failure are the main non-bleeding cause of death in cirrhotic patients with acute GIB [10]. It is thus reasonable to assume that albumin treatment may ameliorate the number of deaths from infection or infection-related organ failures. Although the results showed that more than 40 g of albumin infusion reduce mortality in multivariate regression, we failed to demonstrate the beneficial results of its effect on non-bleeding deaths since most patients dead from hemorrhagic shock. However, based on current evidence supporting its efficacy in these complications, we believe that the infusion of albumin for cirrhosis with GIB is favorable to improve the in-hospital prognosis in those with high-risk for non-bleeding death [14]. Further studies on GIB patients with a high risk of infection and HRS are needed to address this issue.
In this study, the pre-RBC infusion was positively correlated with the occurrence of rebleeding. It is not surprising that patients with rebleeding present severe bleeding at admission, thus require more RBC transfusion. Although the causal relationship between pre-RBC infusion and rebleeding cannot be determined based on current data, this issue has already been described by many well-designed studies [8,18,19]. A limited RBC transfusion reduces the occurrence of rebleeding, the survival benefit was not observed in our study. Its effect on survival was not inconsistent in previous reports probably due to the different enrollment criteria [8,[19][20][21][22]. Nevertheless, the restrictive RBC transfusion is a cost-effective strategy as most studies show that less RBC transfusion does not correlated with poor prognosis.
As a retrospective study, the limitations were inevitable. First, the time of albumin infusion before rebleeding was not evaluated. Normally, the albumin is not recommended in the initial resuscitation of hemorrhagic shock. Whether albumin should be applied in the resuscitation stage or after hemostasis achieved needs to be explored; Second, the average dose of albumin used in our study is low compared with previous studies. The lower dose may be due to the conventional view of its limited effect on GIB, as well as its high cost. Hence, the appropriate dose needs to be evaluated by further studies focused on this issue. Third, the rebleeding was associated with many confounding variables, we cannot exclude the existence of selection bias due to our retrospective design. Although our result has been statistically adjusted by confounders, the applicability of the conclusion requires evaluation in the target population. Last, the 12-day follow-up period was shorter than those described in prior studies. However, as the rebleeding usually occurs within the first two weeks, we believe our analysis was still persuasive [8].

Conclusions
Gastrointestinal bleeding (GIB) is a fatal complication of cirrhosis, especially for those with rupture of varices. The application of endoscopic and radiological interventions has greatly improved the prognosis of patients with acute GIB [9]. In addition to the advances of invasive treatments, the conservative treatments remain an integral part of the management of this medical emergency and is increasingly standardized. The therapeutic value of prophylactic antibiotics and restrict blood transfusion strategy has recently been proved by several clinical studies [5,8]. However, the role of albumin in the management of GIB in patients with decompensated cirrhosis has not yet been assessed. This study demonstrated that albumin infusion was associated with a lower risk of rebleeding and in-hospital mortality in cirrhotic patients.
Our study revealed a rebleeding rate of 10.8% and overall mortality of 6.0%, which were lower than those described in previous studies possibly due to a shorter follow-up period in our study [10,11].
Despite the risk factors that are already known for rebleeding, our study found that the unidentified lesion also increased the risk of rebleeding and in-hospital mortality in cirrhotic patients [5,11].
Current consensus recommended endoscopic intervention for non-variceal upper gastrointestinal bleeding within 24 hours since hemodynamic homeostasis, while the timing of endoscopy in cirrhosis suspected variceal bleeding seems not correlated to the outcomes [12]. Given the unclear urgency of endoscopic intervention and its requirement of technical expertise, endoscopy examinations for suspected variceal bleeding were only delivered after hemodynamic hemostasis in our institute.
However, based on our results, we advocated that appropriate endoscopic therapies for variceal lesion should be given within 5 days as most experts suggested. In consideration of the negative impact of unidentified lesion on the outcomes possible due to the rapid disease deterioration and passive surveillance to invasive treatments, we believe early identification of bleeding lesion is crucial for the management of acute GIB.
Our study aims to explore whether albumin infusion influences the outcomes of cirrhotic patients 12 admitted for GIB. Theoretically, abundant albumin infusion in patients with GIB may cause rebleeding by increased portal pressure in a similar manner of liberal transfusion strategy [13]. Interestingly, we found that albumin infusion reduces the risk of in-hospital rebleeding. Particularly, this beneficial effect was predominately observed in patients with Child-Pugh C class, most of whom had high portal pressure.
In our study, the mean amount of pre-ALB was just 12.3g, which is much lower than those recommended for ascites treatment (1-1.5 g/kg). Also, the dose-dependent effect of albumin in reducing rebleeding does not support the view that a higher dose leads to more bleeding. In fact, a recent study suggests that the portal pressure was not affected after the infusion of albumin (1.5g/kg, every week) [6]. Hence, the concern that albumin infusion increases the risk of rebleeding due to deterioration of portal pressure seems unnecessary. On the other hand, albumin is used in cirrhosis for the treatment of SBP and was widely discussed in the sepsis treatment for its immunomodulatory and anti-inflammatory properties [14,15]. As bacterial infection was identified as a critical determinant of rebleeding in cirrhosis by increasing portal pressure through vasoactive substances [5], the effect of albumin on systemic inflammation provides a potential theoretical basis for its role in reducing bleeding. Moreover, albumin may also control the bleeding risk by promoting the transportation of drugs such as proton pump inhibitor and antibiotics, two essential agents in the management of acute GIB. Indeed, a recent clinical study found that the combination of albumin and antibiotics is superior to antibiotics alone in the control of inflammation [6]. Collectively, the mechanisms involved the benefit of albumin is attribute to its effect on systemic inflammation and pharmacokinetic process, although its other properties may also play a role [16,17].
The association between albumin infusion and in-hospital mortality was also evaluated. In our study, infection, HRS and liver failure are the main non-bleeding cause of death in cirrhotic patients with acute GIB [10]. It is thus reasonable to assume that albumin treatment may ameliorate the number of deaths from infection or infection-related organ failures. Although the results showed that more than 40g of albumin infusion reduce mortality in multivariate regression, we failed to demonstrate the beneficial results of its effect on non-bleeding deaths since most patients dead from hemorrhagic shock. However, based on current evidence supporting its efficacy in these complications, we believe that the infusion of albumin for cirrhosis with GIB is favorable to improve the in-hospital prognosis in those with high-risk for non-bleeding death [14]. Further studies on GIB patients with a high risk of infection and HRS are needed to address this issue.
In this study, the pre-RBC infusion was positively correlated with the occurrence of rebleeding. It is not surprising that patients with rebleeding present severe bleeding at admission, thus require more RBC transfusion. Although the causal relationship between pre-RBC infusion and rebleeding cannot be determined based on current data, this issue has already been described by many well-designed studies [8,18,19]. A limited RBC transfusion reduces the occurrence of rebleeding, the survival benefit was not observed in our study. Its effect on survival was not inconsistent in previous reports probably due to the different enrollment criteria [8,[19][20][21][22]. Nevertheless, the restrictive RBC transfusion is a cost-effective strategy as most studies show that less RBC transfusion does not correlated with poor prognosis.
As a retrospective study, the limitations were inevitable. First, the time of albumin infusion before rebleeding was not evaluated. Normally, the albumin is not recommended in the initial resuscitation of hemorrhagic shock. Whether albumin should be applied in the resuscitation stage or after hemostasis achieved needs to be explored; Second, the average dose of albumin used in our study is low compared with previous studies. The lower dose may be due to the conventional view of its limited effect on GIB, as well as its high cost. Hence, the appropriate dose needs to be evaluated by further studies focused on this issue. Third, the rebleeding was associated with many confounding variables, we cannot exclude the existence of selection bias due to our retrospective design. Although our result has been statistically adjusted by confounders, the applicability of the conclusion requires evaluation in the target population. Last, the 12-day follow-up period was shorter than those described in prior studies. However, as the rebleeding usually occurs within the first two weeks, we believe our analysis was still persuasive [8]. .00 † Total-ALB and total-RBC: the total amount of albumin/RBC infusion during hospitalization; § Pre-ALB and pre-RBC: the amount of albumin infusion before rebleeding occurrence. In patients without bleeding, the pre-ALB and pre-RBC equals the total-ALB and total-RBC.    Multivariate hazard regression of (A) different dose (g) of albumin infusion before rebleeding (total-ALB); (B) different dose (units) of red blood cell (total-RBC) infusion and mortality