Improved survival of cirrhotic patients with variceal bleeding over the decade 2000–2010

Clinics and Research in Hepatology and Gastroenterology (2015) 39, 59—67

Available online at

ScienceDirect www.sciencedirect.com

ORIGINAL ARTICLE

Improved survival of cirrhotic patients with variceal bleeding over the decade 2000—2010 Delphine Vuachet a, Jean-Paul Cervoni a, Lucine Vuitton b, Delphine Weil a, Stavros Dritsas a, Alain Dussaucy c, Stéphane Koch b, Vincent Di Martino a, Thierry Thevenot a,∗ a

Service d’hépatologie et de soins intensifs digestifs, hôpital Jean-Minjoz, 25030 Besanc¸on cedex, France Service de gastro-entérologie, hôpital Jean-Minjoz, 25030 Besanc¸on cedex, France c Département d’information médical, hôpital Jean-Minjoz, 25030 Besanc¸on cedex, France b

Available online 18 August 2014

Summary Background and objective: Advances in the management of variceal bleeding (VB) have been highlighted recently. We aimed at assessing whether changing the management of VB has improved the outcome (mortality and rebleeding rates). Methods: The files of two cohorts (n = 57, 2000—2001 and n = 64, 2008—2009) of patients referred to our university center were reviewed after a cross-searching using two coding systems. Data were recorded during the six months after VB. Results: As compared to 2000—2001, more use of general anesthesia (25.4% vs. 11.1%; P = 0.049), band ligations (96.1% vs. 71.4%; P = 0.001), octreotide (95.3% vs. 80.7%; P = 0.012) and antibiotic prophylaxis (93.8% vs. 82.5%; P = 0.09) were performed in 2008—2009, whereas the number of red-cell units transfused during the hospital stay (4.3 ± 3.2 vs. 7.1 ± 5.7; P = 0.005) decreased. Surprisingly, more than 60% of patients reached the emergency department from home without medical assistance in both periods. In 2008—2009, patients had more comorbidities and no patients underwent early-TIPS but the 6-week mortality rate (24.6% vs.10.9%; P = 0.048) was lower. The 6-week mortality was associated with high MELD score

Abbreviations: CCAM, classification commune des actes médicaux; CdAM, catalogue des actes médicaux; CRP, C-reactive protein; ICD, International Classification of Diseases; MDRD, modification of diet in renal disease; MELD, model of end stage liver disease; TIPS, transjugular intrahepatic portosystemic shunt; UGI, upper gastrointestinal; VB, variceal bleeding. ∗ Corresponding author. Service d’hépatologie et de soins intensifs digestifs, hôpital Jean-Minjoz, 25000 Besanc ¸on, France. Tel.: +33 3 81 66 85 94; fax: +33 3 81 66 84 18. E-mail address: [email protected] (T. Thevenot). http://dx.doi.org/10.1016/j.clinre.2014.06.018 2210-7401/© 2014 Elsevier Masson SAS. All rights reserved.

60

D. Vuachet et al.

(HR = 1.13; 95%CI: 1.08—1.18) and hypovolemic shock (HR = 5.36; 95%CI: 1.96—14.67) at admission. In multivariate analysis adjusted on MELD and comorbidities, the 2008—2009 period (HR: 0.42; 95%CI: 0.20—0.87; P = 0.02) was associated with a lower 6-month mortality rate. Conclusions: Although cirrhotic patients with VB had more comorbidities in 2008—2009 and received no early-TIPS, their prognosis has improved during this last decade concomitantly to a more intensive care and a lower transfusion strategy. © 2014 Elsevier Masson SAS. All rights reserved.

Introduction Until a decade ago, cirrhosis was considered as a progressive and inevitable terminal disease, heralded by the development of ascites, jaundice, encephalopathy and variceal bleeding. Patients presenting with variceal bleeding (VB) have been associated with the highest 1-year mortality rate of 57% [1]. The improvement of hospital care in decompensated cirrhotic patients has led to the significant reduction in the mortality rate over the last decade [2], and one of the most spectacular advances in the field of portal hypertension was the improvement of the management of VB. In a French retrospective study published ten years ago, a 3fold decrease in the in-hospital mortality observed between 1980 and 2000 in cirrhotic patients with VB was attributed to a more general use of pharmalogical and endoscopic therapies together with short-term antibioprophylaxis [3]. Since the publication of this paper, recent advances in the therapeutic strategies of VB have been highlighted. The use of early transjugular intrahepatic portosystemic shunts (TIPS) in patients with high-risk of failing standard therapy and the use of a restrictive red-cell transfusion strategy have been shown to decrease the mortality in such patients [4,5]. Early TIPS, however, is not widely available and its systematic use in all centers is still debated, although recommended by the Baveno V workshop [6]. The benefit of early-TIPS must still be evaluated in competition with other medical measures susceptible to improve survival. We thus compared two cohorts of patients referred to a single center to investigate whether survival still improved over the last decade despite the non-use of early-TIPS. The prognosis of acute VB and its determinants were fully investigated.

Patients and methods Study cohorts This retrospective study was conducted in the 1291-bed University Hospital of Besanc ¸on and compared two cohorts (first cohort: January 2000 to December 2001; second cohort: January 2008 to December 2009) of patients admitted for acute upper gastrointestinal (UGI) bleeding or who bled while being hospitalized for another reason. The files of all patients hospitalized during these two periods with a discharge diagnosis of cirrhosis and acute UGI bleeding were reviewed. All patients who met the following criteria were considered:

• diagnosis of cirrhosis, based on liver histology or on clinical, laboratory or ultrasonographic findings; • hematemesis and/or melaena at admission; • active VB on endoscopy (see Definitions section below) or non-bleeding varices with no other potential source of gastrointestinal bleeding. For the purpose of this study, only VBs were considered. Consequently, patients with portal hypertensive gastropathy, ulcers (whatever the origin), Mallory-Weiss syndrome or sclerotherapy/banding-induced ulcerations were excluded from the analysis.

Data collection Based on the 10th edition of International Classification of Diseases (ICD-10), our Medical Information Department provided the list of all patients with a diagnosis of ‘‘cirrhosis’’ or of ‘‘chronic hepatitis’’ and with acute UGI hemorrhage hospitalized in our Hepatology Department or in the Intensive Care Unit Department, during the two periods. To ensure that all files corresponding to these diagnoses were optimally found, we also reviewed the files of patients who have performed a therapeutic upper endoscopy for hemorrhage. To do so, we used the French catalogue des actes médicaux (CdAM) for the first period, and the classification commune des actes médicaux (CCAM) (this medical classification for the clinical procedures succeeded to CdAM in 2005) for the second period. The medical records of eligible patients were reviewed to obtain demographic, clinical and biological data, endoscopic procedures and follow up data until six months after the first episode of UGI bleeding. Cases of undetermined vital status at the end of the follow-up were resolved by questioning the family physicians or asking about the death certificate at the town hall. Comorbidities were evaluated by the Charlson Index [7], and we calculated the modification of diet in renal disease (MDRD) formula to estimate the glomerular filtration rate [8].

Definitions Active bleeding on admission was defined as objective signs of red or black hematemesis/aspiration on the nasogastric tube seen, or spurting or oozing bleeding at endoscopy. Hypovolemic shock was defined as a systolic pressure below 80 mmHg with signs of hypoperfusion [3].

Variceal bleeding and liver cirrhosis Failure to control bleeding was defined as death or rebleeding (i.e., a new hematemesis or fresh melena) requiring a transfusion of more than two units of red blood cell (RBC) within the five days following the index bleeding. Mortality was defined as death by hypovolemic shock due to massive, uncontrolled bleeding was attributed to the hemorrhage. Death by hypovolemic shock associated with proved or suspected infection was attributed to septic shock. Death by coma, hepatorenal syndrome, or multiorgan failure was attributed to terminal liver failure.

Statistical analyses Qualitative variables are expressed as percentages and compared using Chi2 test or Fischer exact test. Quantitative variables are expressed as mean ± standard deviation (SD) or as median ± interquartile range (IQR) and compared using the Student t test. Actuarial probabilities of transplant-free survival or survival free of rebleeding were calculated using the Kaplan-Meier method and were compared with the logrank test. Variables associated with the considered events in univariate analysis were then entered into a multivariate Cox model using a stepwise procedure with a backward approach. All two-tailed P-values <0.05 were considered significant. Statistical analyses were performed using NCSS 2007 Statistical Program software.

Results Patients’ characteristics In the cohort 2000—2001, 66 cirrhotic patients with an acute episode of UGI hemorrhage were identified. Among them, 57 patients with VB were eligible to enter into the analyses. Using the same selection process, 81 and 64 cirrhotic patients respectively experienced an acute episode of UGI hemorrhage and a VB in 2008—2009 (see the flow chart in Table S1, supplementary material). The source of UGI bleeding displayed similar origins between the two periods (VB: 86.4 vs. 79.0%; P = 0.25; Table S2 in supplementary material). Baseline characteristics of patients are reported in Table 1. Notably, more patients in 20089 received gastrotoxic drugs, anticoagulants or clopidogrel (34.4 vs. 15.8%; P = 0.019) and had also a higher Charlson comorbidity index (6.6 ± 2.2 vs. 5.7 ± 1.9; P = 0.026) than those in cohort 2000—2001. The median length of hospital stay was not significantly different between the two periods [12 days (IQR: 7-22.5)] vs. [nine days (IQR: 6-14)]; P = 0.13.

Characteristics and management of the acute episode of variceal bleeding Characteristics of the acute episode of bleeding, the endoscopic and pharmacological therapies and the prophylaxis used for the acute episode of bleeding are shown in Table 2. Notably, 63.8% of patients did not receive any medical transport to the emergency department. During the period 2008—2009, general anesthesia and endoscopic procedures for the index bleeding were more

61 common, and endotracheal intubations were performed within the Endoscopic Unit in 67% and 44% of cases in 2000—2001 and 2008—2009, respectively. The use of sclerotherapy dropped markedly from 31.4% in 2000—2001 to 2% in 2008—2009, while band ligation (71.4% vs. 96.1%; P = 0.001) and use of octreotide (80.7% vs. 95.3%; P = 0.012) became the methods of choice for the treatment of VB in 2008—2009. The use of balloon tamponade was anecdotal since only five patients in 2000—2001 (two before the endoscopy, one for failure of endoscopic therapy and two for early rebleeding) and two patients in 2008—2009 (for failure of endoscopic therapy) received this procedure. Antibiotic prophylaxis was more routinely used in 2008—2009 (82.5% in 2000—2001 vs. 93.8% in 2008—2009; P = 0.09). In our center, the combination of amoxicillin-clavulanic acid plus quinolones (51%) in 2000—2001, and amoxicillin-clavulanic acid alone (53%) in 2008—2009 were the more frequent antibioprophylaxis strategies. In 2008—2009, practitioners favored a more restrictive strategy for the number of redcell units transfused during the first 24 hours (4.0 ± 2.1 in 2000—2001 vs. 3.0 ± 2.1 in 2008—2009; P = 0.032) or during the entire hospital stay (4.3 ± 3.2 vs. 7.1 ± 5.7; P = 0.005) (7.1 ± 5.7 in 2000—2001 vs. 4.3 ± 3.2 in 2008-9; P = 0.005). A similar finding was made for the transfusions of coagulation factors (24.6% in 2000—2001 vs. 7.8% in 2008-9; P = 0.01). Transjugular intrahepatic portosystemic shunt (TIPS) procedure was performed in four patients in 2000—2001 (two for failure to control bleeding and two for recurrent hemorrhage) and in no patient in 2008—2009. We also observed a 2.5-fold decrease of 5-day failure to control bleeding between the two study periods (15.8 vs. 6.3% in 2008—2009; P = 0.14). Of note, secondary prophylaxis with propranolol in patients surviving at day 5 increased significantly from 54.9% to 83.6% (P = 0.001) in 2000—2001 and 2008—2009, respectively.

Rebleeding at 6 months Overall, 27 episodes of rebleeding occurred, 16 (31.4%) in 2000—2001 and 11 (18.0%; P = 0.10) in 2008—2009 (Table 3). This event rarely concerned Child-Pugh A patients (12.5% in 2000—2001 and 0% in 2008—2009), as compared to Child-Pugh B/C patients (37.5%/50% in 2000—2001 and 72.7%/27.3% in 2008—2009). In univariate analysis performed in the 112 patients surviving after day 5, factors associated with rebleeding were hypovolemic shock (22.2 vs. 7.1%; P = 0.026), ascites (70.4 vs. 47.1%; P = 0.035), band ligation (70.0 vs. 93.5%; P = 0.012), hemoglobin level at admission (7.5 ± 2.3 vs. 8.9 ± 2.1 g/dL; P = 0.0035), MDRD at admission (65.8 ± 30.9 vs. 84.3 ± 35.2 mL/min; P = 0.016) and the number of red-cell units transfused during the first 24 hours (4.4 ± 2.9 vs. 3.0 ± 1.2; P = 0.037) or during the whole hospital stay (8.8 ± 6.3 vs. 4.3 ± 3.4; P = 0.0017). In multivariate analysis (Cox model), independent factors associated with rebleeding at 6-month were the number of red-cell units transfused during the hospital stay (HR: 1.13; 95%CI: 1.06—1.21), ascites (HR: 2.78; 95%CI: 1.12—6.90) and hemoglobin level (HR: 0.75; 95%CI: 0.60—0.94). The 6-month rebleeding curve is presented in Fig. 1.

62

D. Vuachet et al.

Table 1

Clinical and laboratory characteristics of the patients on admission.

Mean age (years) Male gender, n (%) BMI (kg/m2 ) Etiology of cirrhosisa Alcoholic, n (%) Viral, n (%) Otherb , n (%) Child-Pugh score Child Pugh A/B/C, % MELD score Charlson comorbidity index Prothrombin index (%) Albumin (g/L) Bilirubin (mol/L) Ascites, n (%) Encephalopathy, n (%) Platelet count (G/mm3 ) Sodium (mmol/L) Glomerular filtration rate (mL/min) Severe acute alcoholic hepatitisc , n (%) Hepatocellular carcinoma, n (%) Portal vein thrombosis, n (%) Drugs prior to admission, n (%) Gastrotoxic drugs, clopidogrel or anticoagulants Aspirin NSAI drugs Clopidogrel Anticoagulants Proton pump inhibitors Nonselective ␤-blockersd

Whole cohort n = 121

2000—2001 n = 57

2008—2009 n = 64

P

60.9 ± 10.4 94 (77.7%) 25.2 ± 3.8

60.9 ± 12.4 41 (71.9%) 25.6 ± 4.3

61.0 ± 8.5 53 (82.8%) 25.0 ± 3.6

0.96 0.15 0.53

95 (78.5%) 15 (12.4%) 21 (17.6%) 8.9 ± 2.2 16.5/43.8/39.7 18.3 ± 7.4 6.1 ± 2.1 49.8 ± 16.9 26.9 ± 6.5 68.2 ± 9.9 66 (54.6%) 14 (11.6%) 139 ± 72 136.6 ± 5.5 78.2 ± 35.3 38 (31.4%) 19 (15.7%) 9 (7.4%)

47 (82.5%) 6 (10.5%) 6 (10.5%) 9.3 ± 2.4 15.8/36.8/47.4 18.8 ± 8.5 5.7 ± 1.9 51.7 ± 18.1 28.0 ± 6.3 80.7 ± 112.2 35 (61.4%) 7 (12.3%) 124 ± 67 136.5 ± 5.5 79.9 ± 40.2 22 (38.6%) 9 (15.8%) 3 (5.3%)

48 (75.0%) 9 (14.1%) 15 (23.4%) 8.6 ± 2.1 17.2/50.0/32.8 18.0 ± 6.4 6.6 ± 2.2 48.0 ± 15.8 26.0 ± 6.6 57.0 ± 105.0 31 (48.4%) 7 (10.9%) 152 ± 76 136.8 ± 5.6 76.8 ± 30.5 16 (25.0%) 10 (15.6%) 6 (9.4%)

0.32 0.56 0.06 0.09 0.24 0.58 0.026 0.23 0.11 0.23 0.15 0.82 0.036 0.72 0.63 0.11 0.98 0.50

31 (25.6%) 15 (12.4%) 4 (3.3%) 5 (4.1%) 13 (10.7%) 36 (29.8%) 34 (28.1%)

9 (15.8%) 4 (7.0%) 1 (1.8%) 1 (1.8%) 4 (7.0%) 15 (26.3%) 13 (22.8%)

22 (34.4%) 11 (17.2%) 3 (4.7%) 4 (6.3%) 9 (14.1%) 21 (32.8%) 21 (32.8%)

0.019 0.10 0.62 0.36 0.25 0.44 0.23

CRP: C-reactive protein; NSAI drugs: non-steroidal anti-inflammatory drugs. a Patients may have several etiologies of cirrhosis. b Other means nonalcoholic steatohepatitis, autoimmune hepatitis, primary biliary cirrhosis or undefined. c Severe acute alcoholic hepatitis suspected or histologically proven. d Nonselective ␤-blockers included carvedilol, nadolol, propranolol, sotalol.

Survival analysis

Survival free of rebleeding (%)

Overall, we observed a 2-fold decrease in the in-hospital and in the 5-day mortality between the two periods

(Table 3). During the 6-week follow-up, 14 (24.6%) and 7 (10.9%) patients died respectively in cohort 2000—2001 and 2008—2009 (log-rank test, P = 0.048; Fig. 2). The improvement in survival was mainly observed for Child-Pugh C patients. Indeed, 10 and 4 Child-Pugh C patients died in

100 80 60 40

Log-rank Chi-square P value test : 3.14 0.076

20

Days 0 5

Numbers 61 at risk 51

60

120

50

45

180 42

2008-2009

34

29

25

2000-2001

Figure 1 Cumulative probabilities of remaining free of rebleeding after day 5 in the two cohorts.

Figure 2 Cumulative probabilities of 6-week survival in the two cohorts.

Variceal bleeding and liver cirrhosis Table 2

63

Characteristics and management of variceal bleeding.

First variceal bleeding, n (%) Management < 24 hours, n (%) Source of bleedinga : EV/GV/ND, n Red color signs, n (%) Transport to the emergency department without medical assistance Referred from another hospital Patients hospitalized before the bleeding, n (%) Mean arterial pressure (mmHg) Hypovolemic shock, n (%) Active bleeding on admission, n (%) Hemoglobin level on admission (g/dL) Endoscopy delay < 24 hoursa , n (%) General anesthesia with intubationa , n (%) Endoscopic therapya , n (%) Performed on the first procedure, n (%) Ligation/Sclerotherapy/Glue/Combination, n Ligation, n (%) Vasoactive treatmentb , n (%) Octreotide, n (%) Duration (days) Antibiotic use, n (%) Duration (days) Red blood cell transfusion, n (%) On the first 24 hours, n (%)c Nb of units transfused within the first 24 hoursc Nb of patients with pre-transf Hb level between 7—9 g/dL, n (%)c Nb of patients with pre-transf Hb level < 7 g/dL, n (%)c Nb of units transf during the hospital stayc Coagulation factors/platelets transfusion, n (%) Balloon tamponade/TIPS, n (%) Secondary prophylaxis in patients surviving at day 5 Propranolol, n (%) Secondary band ligation, n (%) Propranolol and secondary band ligation, n (%)

Whole cohort n = 121

2000—2001 n = 57

2008—2009 n = 64

P

95 (78.5%) 77 (63.6%) 109/4/8 72 (61.5%) 69 (63.9%)

45 (78.9%) 35 (61.4%) 56/0/1 33 (61.1%) 30 (61.2%)

50 (78.1%) 42 (65.6%) 53/4/7 39 (61.9%) 39 (66.1%)

0.91 0.63 — 0.93 0.60

27 (25.0%) 13 (10.7%) 83.9 ± 17.0 14 (11.6%) 23 (19.0%) 8.6 ± 2.3 100 (85.5%) 22 (18.8%) 86 (73.5%) 64 (74.4%) 74/12/5/5 74 (86.1%) 110 (90.9%) 107 (88.4%) 4.9 ± 3.3 107 (88.4%) 8.0 ± 5.5 95 (78.5%) 82 (86.3%) 3.5 ± 2.1

14 (28.6%) 8 (14.0%) 89.2 ± 17.3 5 (8.8%) 14 (24.6%) 8.9 ± 2.4 45 (83.3%) 6 (11.1%) 35 (64.8%) 27 (77.1%) 25/11/0/1 25 (71.4%) 48 (84.2%) 46 (80.7%) 5.2 ± 4.3 47 (82.5%) 7.4 ± 4.6 46 (80.7%) 38 (82.6%) 4.0 ± 2.1

13 (22.0%) 5 (7.8%) 79.2 ± 15.4 9 (14.1%) 9 (14.1%) 8.4 ± 2.3 55 (87.3%) 16 (25.4%) 51 (81.0%) 37 (72.5%) 49/1/5/4 49 (96.1%) 62 (96.9%) 61 (95.3%) 4.7 ± 2.4 60 (93.8%) 8.4 ± 6.2 49 (76.6%) 44 (89.8%) 3.0 ± 2.1

0.43 0.27 0.001 0.36 0.14 0.24 0.54 0.049 0.048 0.63 — 0.001 0.016 0.012 0.50 0.09 0.36 0.58 0.31 0.032

51 (53.7%)

25 (54.3%)

26 (53.1%)

0.90

39 (41.0%)

17 (37.0%)

22 (44.9%)

0.43

5.7 ± 4.8 7.1 ± 5.7 4.3 ± 3.2 0.005 19 (15.7%)/15 (12.4%) 14 (24.6%)/7 (12.3%) 5 (7.8%)/8 (12.5%) 0.01/0.97 7 (5.8%)/4 (3.3%) 5 (8.8%)/4 (7.0%) 2 (3.1%)/0 0.25/0.046

79 (70.5%) 60 (53.6%) 49 (43.8%)

28 (54.9%) 24 (47.1%) 19 (37.3%)

51 (83.6%) 36 (59.0%) 30 (49.2%)

0.001 0.21 0.12

EV: esophageal varices; GV: gastric varices; ND: not determined. Nb: number. Tranf.: transfused or transfusion. a Three patients in 2000—2001 and one patient in 2008—2009 died before the endoscopy procedure. In these four patients, the presence of oesophageal varices was previously known and we assumed that they were the source of bleeding. These patients were included in the analysis, except for the items regarding the endoscopic treatment. b Two patients received glypressin in 2000—2001 and one received adrenalin in 2008—2009. Gastroesophageal varices type 1 are considered with esophageal varices. GV includes isolated gastric varices type 1 or type 2, and gastroesophageal varices type 2. c Among patients with red blood cell transfusion.

2000—2001 and in 2008—2009, while 4 and 3 Child-Pugh B patients died in 2000—2001 and 2008—2009, respectively; no Child-Pugh A patients died in both cohorts. Variables associated with the 6-week mortality in univariate analysis are shown in Table S3 (supplementary material). MELD score (HR: 1.13; 95%CI: 1.08—1.18; P < 0.0001) and hypovolemic shock at admission (HR: 5.36; 95%CI: 1.96—14.67; P = 0.001) best predicted death at six weeks in a Cox multivariate model. By extending the patient follow-up to six months,

the difference in the mortality rate between the two cohorts decreased slightly from 38.6% (n = 22) to 25.0% (n = 16) (logrank test, P = 0.088;Fig. S1, supplementary material). The improvement of the 6-week or 6-month survival rate was associated with lower death among patients belonging to the Child-Pugh C class (Table 3). In both combined cohorts, causes of death at six months were mainly bleeding-related (32%), end-stage liver failure (29%) and sepsis-related (18%).

64

D. Vuachet et al.

Table 3

Relevant events occurring in the two cohorts.

5-day failure to control bleeding, n (%) Rebleeding between day 5 and week 6a , n (%) In—hospital mortality, n (%) ≤ 24 hours, n (%) ≤ 5 days, n (%) 6—week mortality, n (%) Child A/B/C, n 6—month mortality, n (%) Child A/B/C, n Causes of death at 5 days/6 weeks/6 months, % Variceal bleeding Terminal hepatic failure Sepsis Othersc Unknown Liver transplantation, n (%) a b c

Whole cohort n = 121

2000—2001 n = 57

2008—2009 n = 64

P

13 (10.7%) 27 (24.1%) 22 (18.2%) 5 (4.1%) 9 (7.4%) 21 (17.4%) 0/7/14 38 (31.4%) 1/10/27

9 (15.8%) 16 (31.4%) 14 (24.6%) 3 (5.3%) 6 (10.5%) 14 (24.6%) 0/4/10 22 (38.6%) 1/5/16

4 (6.3%) 11 (18.0%) 8b (12.5%) 2 (3.1%) 3 (4.7%) 7 (10.9%) 0/3/4 16 (25.0%) 0/5/11

0.14 0.10 0.09 0.67 0.30 0.048 0.64 0.11 0.61

77.8/47.6/31.6 11.1/28.6/28.9 11.1/14.3/18.4 0/4.8/7.9 0/4.8/13.2 4 (3.5%)

83.3/50.0/36.4 0/21.4/27.3 16.7/14.3/9.1 0/7.1/13.6 0/7.1/13.6 3 (5.3%)

66.7/42.9/25.0 33.3/42.9/31.3 0/14.3/31.3 0/0/0 0/0/12.5 1 (1.7%)

— — — — — 0.36

In the whole cohort, 112 patients survived after day 5, 51 in 2000—2001 and 61 in 2008—2009. One patient died after 6 weeks during the same hospitalization. Other causes of death in 2000—2001: early complication of liver transplantation (n = 1) or death from cardiac cause (n = 2).

Compared to survivors (n = 83) in both combined cohorts, the deceased patients (n = 38) at six months had mainly: • significantly higher Child-Pugh (10.4 ± 2.2 vs. 8.2 ± 1.9; P < 0.0001) and MELD scores (23.3 ± 9.3 vs. 16.0.±4.9; P < 0.0001), Charlson comorbidity index (6.7 ± 2.2 vs. 5.9 ± 2.0; P = 0.042), age (63.8 ± 12.6 vs. 59.7 ± 9.0 years; P = 0.045), CRP [24 (IQR: 9—44)] vs. 7 (IQR: 3—15) mg/L; P = 0.0007; • significantly lower natremia (135.1 ± 6.9 vs. 137.3 ± 4.6 mmol/L; P = 0.037) and MDRD (62.8 ± 31.6 vs. 85.3 ± 34.7 mL/min; P = 0.0009). Deceased patients at six months displayed: • more frequently a first episode of VB (94.7% vs. 71.1%; P = 0.0035), active bleeding (31.6% vs. 13.3%; P = 0.017), hypovolemic shock (21.1% vs. 7.2%; P = 0.027), rebleeding after day 5 (37.9% vs. 19.3%; P = 0.043), ascites (87% vs. 40%; P < 0.0001), severe acute alcoholic hepatitis (53% vs. 22%; P = 0.0007), bacterial infection (52.6% vs. 19.3%; P = 0.0002); • a less intensive use of secondary prophylaxis (for nonselective beta-blockers: 44.8% vs. 79.5%; P = 0.0004; for band ligation: 31.0% vs. 61.4%; P = 0.005). In multivariate analysis, the 6-month mortality of the whole population was best predicted by the MELD score (HR: 1.11; 95%CI: 1.06—1.16; P < 0.0001), ascites (HR: 7.18; 95%CI: 2.45—21.00; P = 0.0003), hypovolemic shock (HR: 5.37; 95%CI: 2.24—12.87; P = 0.0002) and CRP level (HR: 1.02; 95%CI: 1.01—1.04; P = 0.0028). The impact of period (HR: 0.42; 95%CI: 0.20—0.87; P = 0.02) on 6-month mortality was also significant in a multivariate model adjusted on MELD (HR: 1.12; 95%CI: 1.07—1.17; P < 0.001) and

Charlson comorbidity index (HR: 1.23; 95%CI: 1.07—1.41; P = 0.003).

Discussion In this retrospective study performed in a French university hospital during the last decade 2000—2010, we were looking for the relevant changes in the management of VB and their potential benefits for cirrhotic patients. Variceal bleeding accounted for 82.3% of all UGI hemorrhage, which was in line with the generally agreed frequency of 70—80% of all UGI bleeding in cirrhotic patients [9,10]. We found that the treatment of these bleeding patients was more effective in 2008—2009, resulting in a marked decrease in the 6-week mortality rate (24.6% vs. 10.9%; P = 0.048), which was lower than the 14.5% mortality rate observed by Carbonell et al. in 2000 [3]. The 6-month rebleeding rate also decreased, albeit non-significantly. These results are particularly relevant because they concerned more fragile cirrhotic patients with substantial comorbidities (especially cardiovascular disease, peripheric vascular disease and strokes: 10.5 vs. 25%, P = 0.04) in 2008—2009, as reflected by a higher Charlson comorbidity index. In agreement with these findings, a higher number of patients with non-alcoholic steatohepatitis (14.1 vs. 1.8%; P = 0.02), and a greater consumption of gastrotoxic drugs, antiplatelets or anticoagulants (34.4 vs. 15.8%; P = 0.019) were observed in 2008—2009. One surprising result was the 64% proportion of patients who reached the emergency department from home without medical assistance in both periods. It means that these patients coming to the hospital on their own could not benefit from the early administration of vasoactive drugs by the mobile medical emergency assistance, a procedure, which has proved to significantly reduce the mortality [11]. In our study, only one patient (1.6%) in 2008—2009 received a

Variceal bleeding and liver cirrhosis prehospital vasoactive therapy while nearly 35% of the patients reported by Carbonnel et al. in their last cohort (year 2000) received this procedure in the Paris area [3]. Efforts in the prehospital management of VB should be endorsed by scientific society of emergency physicians to extend this practice. The improvement of survival and the decrease of rebleeding rates observed in our study were in line with what has been reported during the past two decades. Indeed, vasoactive therapy was administered more intensively and was also more frequently combined with endoscopic treatment. Terlipressine used to lower the portal pressure has been completely abandoned in 2008—2009 in our center since octreotide has fewer side effects, and has a proved effectiveness to achieve hemostasis when combined to endoscopic therapy, and it was the choice of our pharmacy department to provide the less expensive care. [12]. Sclerotherapy has progressively declined over this last decade since only one procedure (vs. 11 procedures during the first period) was performed in 2008—2009; it has been replaced by band ligation, which became the treatment of choice due to a more effective control of VB, a lower rebleeding and mortality rates than sclerotherapy [13]. It is worth of note that band ligation was frequently performed in 2008—2009 even though we recorded less active bleeding during this period (14.1% vs. 24.6% in 2000—2001); practitioners were probably less reluctant to perform band ligation even if bleeding was interrupted at the time of the endoscopic procedure. This illustrates that practitioners were able to identify the varices as the origin of bleeding more often, even when it had stopped at the time of endoscopy. As a consequence, we noticed that secondary prophylaxis with propranolol, a well-known procedure to reduce rebleeding and mortality rates [14,15], dramatically increased from 54.9% in 2000—2001 to 83.6% in 2008—2009. Endotracheal intubation prior to endoscopy was more widely implemented in 2008—2009 when there were any concerns about the airway safety. Practitioners working in our ICU were less reluctant to accept these intubated fragile patients in 2008—2009 probably due to a close collaboration with hepato-gastroenterologists and they acknowledged that a more intensive care of VB may prevent rare fatal episodes of massive aspiration [16]. As demonstrated in a recent French survey on the use of antibiotic prophylaxis in cirrhotic patients [17], we note that antibiotics use became an integral part of the management of UGI hemorrhage during the period 2008—2009, since 94% of cirrhotic patients with VB received this prophylaxis versus 82% only in the first period. Such use of antibiotic prophylaxis in UGI bleeding was in total agreement with international guidelines [18,19]. It is justified by the prevention of bacterial infections, the decrease of the rebleeding rate and the improvement of survival [20—22]. Between the two periods, the proportion of transfused patients remained stable (78%), but the number of units transfused during the first 24 hours or during the hospital stay was significantly lower in 2008—2009. This restrictive management of blood transfusions during the hospital stay translated into a trend towards a decrease in the rebleeding rate after day 5. In multivariate analysis, three main factors were associated with a decrease in the 6-month rebleeding: the number of red-cell units transfused during

65 the hospital stay, ascites and hemoglobin level on admission. It is well-known that blood volume replacement may induce rebound increases in portal pressure and rebleeding, even in the absence of over transfusion [23]. Moreover, blood transfusion has an immunosuppressive effect leading to an increase risk of infection [24]. Long storage time of transfused blood may contribute to VB or other complications of cirrhosis, since red-cell function and viability after transfusion may be reduced due to a decreased red-cell deformability, a reduced capacity of oxygen delivery, a reduction in the concentration of nitric oxide and accumulation of proinflammatory bioactive substances [25]. Villanueva demonstrated that a restrictive transfusion strategy, i.e., starting transfusion when the hemoglobin level falls below 7 g/dL, significantly decreases the risk of further bleeding, the need for rescue therapy and the mortality rate in patients with acute UGI bleeding [5]. In 2008—2009, the use of coagulation factors also significantly decreased, reflecting the growing knowledge that cirrhotic patients could have a hypercoagulable status [26]. The two other factors associated with a decrease in the 6-month rebleeding, ascites and a low hemoglobin level, reflect the high degree of portal hypertension with a higher severity of the bleeding and consequently a higher risk of rebleeding. The use of early TIPS (within up to 72 hours after admission) in patients at ‘‘high risk’’ of treatment failure (Child-Pugh C < 14 or Child-Pugh B with active bleeding) is associated with a reduction in failure to control bleeding, a lower incidence of rebleeding and a decreased mortality rate [4]. Although this practice became a major step forward in the emergency treatment of bleeding varices [6], no early TIPS was inserted in 2008—2009. This fact allows us to better evaluate the improvement on survival and rebleeding associated with the more appropriate medical management of VB in 2008—2009. Considering ‘‘high-risk’’ patients in 2008—2009, 20 (31.2%) would have received early TIPS, but the 6-week mortality did not differ significantly between these 20 patients and the 44 other patients (10.0 vs. 11.4%). Nevertheless, the 6-week mortality reported in the earlyTIPS group of García-Pagán’s study [4] was lower (only 3%) than that observed (10%) in our 20 ‘‘high-risk’’ patients receiving a pharmacological and endoscopic treatment. Further evaluations using recent cohorts of high-risk patients are needed to determine the risk reduction in the mortality rate between those managed with early TIPS placement and those with a standard therapy for acute bleeding varices. The most powerful independent predictors of mortality were represented herein by MELD score, ascites, hypovolemic shock and CRP. All these variables but CRP have been reported to impact the prognosis of cirrhotic patients with VB [3,27,28]. The MELD score was shown to predict more accurately the short and long-term mortality than the Child-Pugh score [27]. We have recently reported that high CRP levels predicted short-term mortality in severe cirrhotic patients [29] and was probably the consequence of events, such as bacterial translocation, that temporarily worsen liver function. We acknowledge that our study suffers from several limitations. First, because this study was retrospective, concerns can be raised regarding the completeness and accuracy of the collected data. To ensure the exhaustiveness, a cross-search was performed using two coding systems

66 (ICD-10 and CdAM or CCAM). Second, because the sample size of our two cohorts can be considered as limited. However, it is in the range of previous published series where the number of patients included per year is compared [9,27,30]; furthermore, the substantial number of events (rebleeding and death) at six months allowed us to investigate through multivariate analysis the therapeutic changes observed during this last decade that significantly impact on survival. Finally, the non-use of early TIPS in our second period allows us to better evaluate the importance of the appropriate use of medical resources, available anywhere, to treat VB. We think these data harvested in a recent cohort of patients who did not undergo early TIPS will be of particular interest since the results of future similar studies including a subset of patients receiving early TIPS will be difficult to interpret. In conclusion, this study suggests that the prognosis of VB in cirrhotic patients has further improved during this last decade 2000—2010, leading herein to a decrease in the 5-day failure to control bleeding, in the rebleeding after day 5 and in the mortality rate. Such findings were probably the result of a marked improvement in the general management of VB consisting in an increased use of prophylactic antibiotics and vasoactive drugs, concurrently with a more restrictive transfusion strategy. Band ligation, which has replaced sclerotherapy, was also a major step forward. Nevertheless, substantial efforts still could be made in the prehospital management of these patients by starting vasoactive drugs as soon as VB is suspected.

Disclosure of interest The authors declare that they have no conflicts of interest concerning this article.

Acknowledgment We wish to thank Miss Vuachet Marjolaine for the correction of this manuscript without funding sources.

Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.clinre.2014.06.018.

References [1] D’Amico G, Garcia-Tsao G, Pagliaro L. Natural history and prognostic indicators of survival in cirrhosis: a systematic review of 118 studies. J Hepatol 2006;44:217—31. [2] Vergara M, Clèries M, Vela E, et al. Hospital mortality over time in patients with specific complications of cirrhosis. Liver Intern 2013;33:828—33. [3] Carbonell N, Pauwels A, Serfaty L, et al. Improved survival after variceal bleeding in patients with cirrhosis over the past two decades. Hepatology 2004;40:652—9. [4] García-Pagán JC, Caca K, Bureau C, et al. Early use of TIPS in patients with cirrhosis and variceal bleeding. N Engl J Med 2010;362:2370—9.

D. Vuachet et al. [5] Villanueva C, Colomo A, Bosch A, et al. Transfusion strategies for acute upper gastrointestinal bleeding. N Engl J Med 2013;368:11—21. [6] De Franchis R. Revising consensus in portal hypertension: report of the Baveno V consensus workshop on methodology of diagnosis and therapy in portal hypertension. J Hepatol 2010;53:762—8. [7] Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987;40: 373—83. [8] Levey AS, Bosch JP, Lewis JB, et al. A more accurate method to estimate glomerular filtration rate from serum creatinine: a new prediction equation. Ann Intern Med 1999;130: 461—70. [9] D’Amico G, De Franchis R. Upper digestive bleeding in cirrhosis. Post-therapeutic outcome and prognostic indicators. Hepatology 2003;38:599—612. [10] Thomopoulos K, Theocharis G, Mimidis K, et al. Improved survival of patients presenting with acute variceal bleeding: prognostic indicators of short- and long-term mortality. Dig Liver Dis 2006;38:899—904. [11] Levacher S. Early administration of terlipressin plus glyceryl trinitrate to control active upper gastrointestinal bleeding in cirrhotic patients. Lancet 1995;346:865—8. [12] Ba˜ nares R, Albillos A, Rincón D, et al. Endoscopic treatment versus endoscopic plus pharmacologic treatment for acute variceal bleeding: a meta-analysis. Hepatology 2002;35: 609—15. [13] Villanueva C, Piqueras M, Aracil C, et al. A randomized controlled trial comparing ligation and sclerotherapy as emergency endoscopic treatment added to somatostatin in acute variceal bleeding. J Hepatol 2006;45:560—7. [14] Bernard B, Lebrec D, Mathurin P, et al. Beta-adrenergic antagonists in the prevention of gastrointestinal rebleeding in patients with cirrhosis: a meta-analysis. Hepatology 1997;25:63—70. [15] D’Amico G, Pagliaro L, Bosch J. Pharmacological treatment of portal hypertension: an evidence-based approach. Semin Liver Dis 1999;19:475—505. [16] Rudolph SJ, Landsverk BK, Freeman ML. Endotracheal intubation for airway protection during endoscopy for severe upper GI hemorrhage. Gastrointest Endosc 2003;57:58—61. [17] Thevenot T, Degand T, Grelat N, et al. A French national survey on the use of antibiotic prophylaxis in cirrhotic patients. Liver Int 2013;33:389—97. [18] Ginès P, Angeli P, Lenz K, et al. EASL clinical practice guidelines on the management of ascites, spontaneous bacterial peritonitis, and hepatorenal syndrome in cirrhosis. J Hepatol 2010;53:397—417. [19] De Franchis R. Updating consensus in portal hypertension: report of the Baveno III Consensus Workshop on definitions, methodology and therapeutic strategies in portal hypertension. J Hepatol 2000;33:846—52. [20] Hou MC, Lin HC, Liu TT, et al. Antibiotic prophylaxis after endoscopic therapy prevents rebleeding in acute variceal hemorrhage: a randomized trial. Hepatology 2004;39:746—53. [21] Bernard B, Grangé J-D, Nguyen-Khac E, et al. Antibiotic prophylaxis for the prevention of bacterial infections in cirrhotic patients with gastrointestinal bleeding: a meta-analysis. Hepatology 1999;29:1655—61. [22] Chavez-Tapia NC, Barrientos-Gutierrez T, Tellez-Avila F, et al. Meta-analysis: antibiotic prophylaxis for cirrhotic patients with upper gastrointestinal bleeding — an updated Cochrane review. Aliment Pharmacol Ther 2011;34:509—18. [23] Kravetz D, Sikuler E, Groszmann RJ. Splanchnic and systemic hemodynamics in portal hypertensive rats during hemorrhage and blood volume restitution. Gastroenterology 1986;90:1232—40.

Variceal bleeding and liver cirrhosis [24] Vamvakas EC, Blajchman MA. Transfusion-related immunomodulation (TRIM): an update. Blood Rev 2007;21: 327—48. [25] Koch CG, Li L, Sessler DI, Figueroa P, et al. Duration of red-cell storage and complications after cardiac surgery. N Engl J Med 2008;358:1229—39. [26] Tripodi A, Anstee QM, Sogaard KK, et al. Hypercoagulability in cirrhosis: causes and consequences. J Thromb Haemost 2011;9:1713—23. [27] Sempere L, Palazón JM, Sánchez-Payá J, et al. Assessing the short- and long-term prognosis of patients with cirrhosis

67 and acute variceal bleeding. Rev Espa˜ nola Enfermedades Dig 2009;101:236—48. [28] El-Serag HB, Everhart JE. Improved survival after variceal hemorrhage over an 11-year period in the Department of Veterans Affairs. Am J Gastroenterol 2000;95:3566—73. [29] Cervoni J-P, Thévenot T, Weil D, et al. C-reactive protein predicts short-term mortality in patients with cirrhosis. J Hepatol 2012;56:1299—304. [30] Cheung J, Soo I, Bastiampillai R, et al. Urgent vs. non-urgent endoscopy in stable acute variceal bleeding. Am J Gastroenterol 2009;104:1125—9.