- Email: [email protected]
Prognostic factors in patients with refractory ascites treated by transjugular intrahepatic porto-systemic shunt: From the liver to the kidney
Digestive and Liver Disease 46 (2014) 1001–1007
Contents lists available at ScienceDirect
Digestive and Liver Disease journal homepage: www.elsevier.com/locate/dld
Liver, Pancreas and Biliary Tract
Prognostic factors in patients with refractory ascites treated by transjugular intrahepatic porto-systemic shunt: From the liver to the kidney Benjamin Hamel a , Olivier Guillaud a , Sabine Roman a,b , Mélanie Vallin a , Frank Pilleul a,b , Pierre-Jean Valette a,b , Luc Henry a , Aymeric Guibal a , Franc¸ois Mion a,b , Jérôme Dumortier a,b,∗ a b
Department of Digestive Diseases, Hospices Civils de Lyon, Edouard Herriot Hospital, France Claude Bernard Lyon 1 University, Lyon, France
a r t i c l e
i n f o
Article history: Received 10 December 2013 Accepted 30 June 2014 Available online 2 August 2014 Keywords: Ascitis Cirrhosis Survival TIPS
a b s t r a c t Background: The aim of this retrospective study was to evaluate the prognostic value of different scores (including Child–Pugh and Model for End Stage Liver Diseases) in cirrhotic patients treated with transjugular intrahepatic porto-systemic shunt for refractory ascites. Methods: Overall, 111 patients with transjugular intrahepatic porto-systemic shunt insertion between January 1998 and July 2012 were included. Results: Survival rates (without transplantation) were 82.0% at 3 months, and 59.4% at 1 year. In addition to standard parameters, a new simple classification based on platelet count and glomerular filtration rate showed strong prognostic ability and could distinguish 3 groups of patients (Log-rank test, p < 0.001): a “good-prognosis” group with platelet counts above 125 × 109 L−1 and a glomerular filtration rate above 90 mL/min (1-year survival rate 92%), a “poor-prognosis” group with platelet counts below 125 × 109 L−1 and a glomerular filtration rate below 90 mL/min (1-year survival rate 34.8%), and an “intermediateprognosis” group (1-year survival rate 58.2%). Multivariate analysis showed a hazard ratio of 6.34 for the intermediate class and of 12.623 for the high class. Conclusions: A new and simple classification including platelet count and glomerular filtration rate is highly predictive of survival in patients with refractory ascites treated with transjugular intrahepatic porto-systemic shunt and could be used to select patients for this procedure. © 2014 Editrice Gastroenterologica Italiana S.r.l. Published by Elsevier Ltd. All rights reserved.
1. Introduction Ascites is the most frequent complication of liver cirrhosis, especially if compared to jaundice, gastrointestinal bleeding, or hepatic encephalopathy [1]. Approximately 50% of cirrhotic patients present with ascites during 10 years of observation [1]. The prognosis for patients with newly established ascites is a 1 and 5-year survival of 85% and 56%, respectively [2]. Life-threatening complications of refractory ascites include spontaneous bacterial peritonitis, abdominal wall hernia (with the risk of incarceration, skin ulceration, or rupture), hepatic hydrothorax, denutrition, and impaired renal function leading to hepatorenal syndrome (HRS).
∗ Corresponding author at: Pavillon Hbis, Hôpital Edouard Herriot, 69437 Lyon Cedex 03, France. Tel.: +33 4 72 11 01 11; fax: +33 4 72 11 01 47. E-mail address: [email protected] (J. Dumortier).
In addition, chronic ascites also strongly affects quality of life [3]. Approximately 90% of cases could be controlled by a combination of optimal diuretic treatment and dietetic rules [1]. The treatment options for diuretic-resistant ascites include therapeutic paracentesis with intravenous albumin infusion, peritoneo-venous shunting, transjugular intrahepatic porto-systemic shunt (TIPS) and liver transplantation (LT) [4]. LT is the only curative treatment, but cannot be largely used because of organ scarcity and extrahepatic contraindications. Several large-scale, randomized, controlled trials of TIPS compared with large-volume paracentesis have demonstrated that TIPS significantly improves the transplant-free survival of cirrhotic patients with refractory ascites [5]. Nevertheless, there is a significant risk of mortality in the first year following TIPS insertion, especially due to worsening of liver function [6]. Therefore, it has been a challenging issue for physicians to elaborate reliable tools for predicting the outcome of patients with refractory ascites and who
http://dx.doi.org/10.1016/j.dld.2014.06.013 1590-8658/© 2014 Editrice Gastroenterologica Italiana S.r.l. Published by Elsevier Ltd. All rights reserved.
1002
B. Hamel et al. / Digestive and Liver Disease 46 (2014) 1001–1007
are candidates for TIPS placement. The Model for End Stage Liver Diseases (MELD) score was initially specifically designed to predict the 3-month mortality risk of patients undergoing elective TIPS [7]. During the past 10 years, in addition to MELD, the prognostic value of several scores has been investigated in this field, including Child–Pugh score, creatinine-modified Child–Pugh score I and II, MELD-Na, iMELD, MESO, and REFIT-MELD. Recently, it has been suggested that a simple score, calculated from bilirubin level and platelet count, could be very efficient for the prediction of survival for patients with refractory ascites treated by TIPS [8]. The aims of the present retrospective study were [1] to evaluate the reliability of different scores (including the standard Child–Pugh and MELD scores and the simple “bilirubin-platelets” score) for the prediction of mortality after TIPS insertion in cirrhotic patients presenting refractory ascites, and [2] to investigate the ability of the same scores to predict ascites control. 2. Patients and methods 2.1. Study population One hundred and eleven consecutive patients who had undergone TIPS placement for refractory ascites complicating liver cirrhosis were included in our centre (Edouard Herriot Hospital, Lyon, France) between January 1998 and July 2012. Refractory ascites was defined as ascites that does not respond to sodium restriction and high-dose diuretic treatment (400 mg/d spironolactone and 160 mg/d furosemide) or that reoccurs rapidly after therapeutic paracentesis or associated with diuretic-induced complications, as proposed by the International Ascites Club [9]. Patients’ demographic characteristics (gender, age) and information on the type of cirrhosis were available for all patients. Baseline biological patients’ characteristics were recorded within one week before TIPS placement and were available for all the patients. Glomerular filtration rate (GFR) was calculated by using the MDRD (Modification of Diet in Renal Disease) simplified formula as follows: 186 × (creatinine/88.4) − 1.154 × age − 0.203 × (0.742 if female gender) × (1.212 if African descent) [10]. 2.2. Calculation of scores The Child–Pugh score was calculated in accordance to Pugh et al. [11]. The other scores were calculated as follows: MELD = 9.6 × loge [creatinine (mg/dL)] + 3.8 × loge [bilirubin (mg/dL)] + 11.2 × loge (INR) + 6.43 [6] [INR: international normalized ratio] MELD-Na = MELD + 1.59 × [135 − Na (mmol/L)]; Na range = 125–140 mmol/L [12] iMELD = MELD + [0.3 × age (years)] − [0.7 × Na (mmol/L) + 100] [13] MESO = [MELD/Na (mmol/L)] × 100 [14] REFIT MELD = 4.082 × log bilirubin (mg/dL) + 8.485 × log creatinine (mg/dL) + 10.671 × log INR + 7.432 [15] REFIT MELD Na = 4.258 × log bilirubin (mg/dL) + 6.792 × log (mg/dL) + 8.290 × log INR + 0.652 log creatinine (140 − Na) − 0.194 × 140 − Na × bilirubin + 6.327 [15] Modified creatinine Child–Pugh score-I (range: 5–19) was derived from the original CP score by adding 0 points for creatinine <1.3 mg/dL and 4 points for creatinine > or = 1.3 mg/dL [16]. Modified creatinine Child–Pugh score-II (range: 5–19) was derived from the original CP score by adding 0 points for
creatinine <1.3 mg/dL, 2 points for creatinine 1.3–1.8 mg/dL and 4 points for creatinine >1.8 mg/dL [15]. 2.3. Patients’ outcome All patients were followed after TIPS insertion. For the analysis of TIPS effectiveness, ascites was considered controlled if patients did not need paracenteses at 6 months. The presence of ascites was assessed and recorded after TIPS revision when this was necessary. For the survival analysis, the follow-up time was defined from the TIPS placement up to death, LT, or last visit (patients who underwent LT were censored as dead). Follow-up was complete through June 30, 2013. 2.4. Statistical analysis Quantitative variables were described using mean, median, range, and standard deviation (SD). Qualitative values were tabulated and percentages were calculated. Quantitative variables were compared using the Student’s t-test or Mann–Whitney test. Qualitative variables were compared using the 2 test. The concordance statistic was calculated both for the scores and tests predicting mortality (at 3 and 12 months), and for those predicting TIPS efficacy (at 6 months). The concordance statistic, c, is identical to the area under the receiver operating characteristic (ROC) curve. The c-statistic varies between 0.5 and 1.0 for sensible models; the higher the value, the better. Only the results of c-statistics above 0.5 were considered as statistically significant. Kaplan–Meier curves were used to represent survival according to the levels of scores and tests. The purpose of the univariate analysis was to select potentially explanatory variables. The continuous variables were dichotomised, by splitting at their median or upper normal value or at previously proposed values, and then compared. A Cox regression model was used to test the significance of these scores and tests as predictors of survival. All tests with a p < 0.1 at univariate analysis were included in the Cox Model. Statistical tests were considered as significant if p < 0.05. All analyses were performed using SPSS (version 18). 3. Results 3.1. Patients’ characteristics The main characteristics of the patients included in the study group are summarized in Table 1. From the whole cohort of 111 patients, 36 subjects were considered to present resistant ascites (no response to high doses of diuretics) and the other 75 presented intractable ascites (with impossibility to give diuretics because of severe side effects or complications). Most of the patients were male, had alcoholic cirrhosis (78.4%), and had a mean age of 56.8 ± 10.7 years. The mean Child–Pugh score was 9 ± 1.45 and mean MELD score was 14 ± 6. No TIPS insertion failure was recorded during the study period for this specific indication. 3.2. TIPS effectiveness After TIPS placement, 24.3% of patients (27/111) underwent a TIPS revision for stenosis and/or thrombosis. The presence of ascites was evaluated and recorded after TIPS revision when this was necessary. The persistence rate of ascites was 45.9% (51/111) at 6 months. None of the biological or clinical parameters considered (including stent type, covered or uncovered) were significantly different when comparing patients with or without ascites persistence. Only
B. Hamel et al. / Digestive and Liver Disease 46 (2014) 1001–1007
1003
Table 1 Characteristics of patients before transjugular intrahepatic portosystemic shunt placement.
Age (years) Male Cause of cirrhosis Alcohol Viral hepatitis Other TIPS Uncovered stent Covered stent Prothrombin index (%) INR Serum bilirubin (mol/L) Serum sodium (mmol/L) Platelet count (×109 L–1 ) Serum creatinine (mol/L) GFR (MDRD) Serum albumin (g/L) (n = 110) Child–Pugh score Child–Pugh class B/C Modified creatinine Child–Pugh score I Modified creatinine Child–Pugh score II MELD score MELD-Na score Refit MELD Refit MELD-Na MESO iMELD
Mean
Median [range]
56.76 ± 11 80 (72.1%)
56 [24–83]
87 (78.4%) 14 (12.6%) 10 (9%) 31 (27.9%) 80 (72.1%) 58.77 ± 16 1.56 ± 0.5 40.84 ± 61
58 [2–100] 1.45 [1–3.8] 25 [5–487]
135.03 ± 5
136 [119–149]
147.41 ± 91
130 [30–586]
89.42 ± 49
74 [27–316]
94.2 ± 43.2 30.47 ± 6
86.1 [17.7–250.8] 31 [10–43]
9±1 79 (71.2%)/32 (28.8%) 10 ± 2
9 [7–13]
9±2
9 [7–15]
14 ± 6 17 ± 6 14 ± 6 13 ± 6 11 ± 4 37 ± 7
13.1 [6.4–38.8] 16.8 [6–39.2] 13.24 [5.5–37.6] 12.96 [4.8–35.2] 9.77 [4.3–29.1] 35.5 [22.7–60.8]
9 [7–15]
INR, index normalized ratio; MDRD, Modification of Diet in Renal Disease; MELD, Model for End Stage Liver diseases; MESO, MELD to sodium index.
the difference in the initial creatinine level was close to statistical significance: 98.5 ± 58 mol/L for patients with ascites vs. 81.7 ± 8 mol/L for patients without ascites (p = 0.05). 3.3. Survival After TIPS insertion, overall actuarial survival rates (without LT) were 82% at 3 months, 72.9% at 6 months, 59.4% at 1 year, and 38.5% at 5 years: 19 patients died or underwent LT at 3 months (4 LT and 15 deaths), and 43 patients died or underwent LT at 1 year (16 LT and 27 deaths). 3.4. Prognostic factors Table 2 lists the results of univariate survival analysis according to clinical and biological variables (Table 2A) and scores (Table 2B). All the following parameters were associated with a significantly better 3-month survival: higher prothrombin level, lower INR, lower Child–Pugh score, lower modified creatinine Child Pugh scores I and II, lower MELD score, lower MELD-Na score, lower Refit MELD, lower MESO score, and lower iMELD score. Similarly, higher TP level, higher MDRD, lower creatinine, lower MELD score, lower Refit MELD and Refit MELD-Na, and lower iMELD were all significantly associated with a better 1-year survival. When survival was analyzed according to gender, 83.8% of men and 87.1% of women were found to be alive at 3 months (p = 0.66), and 61.3% of men and 61.3% of women were alive at 1 year (p = 0.997). Among patients with alcohol-related cirrhosis, 75/87 (86.2%) were alive at 3 months compared with 19/24 (79.2%) of those with non-alcoholic cirrhosis (p = 0.397). At 1 year, 58/87
Fig. 1. Actuarial rates of survival according to platelet-MDRD classification (Kaplan–Meier curves, log-rank p value < 0.001).
(66.7%) of patients with alcoholic cirrhosis and 10/24 (41.7%) of those with non-alcoholic cirrhosis were alive (p = 0.026). Table 3 shows the prognostic ability (expressed as c-statistics) of the biological parameters and scores for the two periods of followup (3 months and 1 year). While MELD and Refit MELD scores had the highest prognostic ability for 3-month survival, creatinine and GFR had the highest prognostic ability for 1-year survival. The ability of both clinical and simple biological parameters to predict actuarial 1-year survival was evaluated according to the Log-rank test (Table 4A). Platelet count, alcoholic origin of cirrhosis, and MDRD were associated with a significantly better survival. Multivariate analysis disclosed that variables with a significant prognostic ability were a platelet count below 125 × 109 L−1 [HR 2.096 (1.129–3.891), p = 0.019], an MDRD above 90 mL/min [HR 3.234 (1.611–6.495), p = 0.001], and the alcoholic origin of cirrhosis [HR 0.506 (0.265–0.966), p = 0.039]. The ability of the scores to predict actuarial 1-year survival was evaluated with the Log-rank test (Table 4B). The following parameters were all associated with a significantly worse survival: Child–Pugh class C, MELD score above 18, MELD score above 13, Refit MELD score above 13, MESO score greater than 10, and iMELD score above 35.5. The combination of a bilirubin level below 50 mol/L and a platelet count above 75 × 109 L−1 was not predictive of survival. Nevertheless, the combination of a bilirubin level above 50 mol/L and a platelet count below 75 × 109 L−1 was predictive of survival: the 1-year survival rate of patients with platelet count above 75 × 109 L−1 or a bilirubin level lower than 50 mol/L was 62.8%, and was 37.5% for patients with platelet count below 75 × 109 L−1 and bilirubin level above 50 mol/L (p = 0.019). Since prognosis was significantly and independently modified in our cohort according to platelet count and MDRD estimation of GFR (from multivariate analysis), we developed a new and simple classification, the PLT-MDRD. This classification divided patients into three groups (Fig. 1): the “good-prognosis” group (group A, n = 25), with a platelet count above 125 × 109 L−1 and an MDRD over 90 mL/min, the “intermediate-prognosis” group (group B, n = 63), with a platelet count above 125 × 109 L−1 and an MDRD below 90 mL/min or platelet count below 125 × 109 L−1 and an MDRD above 90 mL/min, and the “poor-prognosis” group (group C, n = 23), with a platelet count below 125 × 109 L−1 and an MDRD below 90 mL/min. The 1-year actuarial survival rate was 92% for group A, 58.2% for group B, and 34.8% for group C
1004
B. Hamel et al. / Digestive and Liver Disease 46 (2014) 1001–1007
Table 2A Univariate analysis of survival according to clinical and biological data.
Age (years) Prothrombin rate INR Serum bilirubin (mol/L) Serum sodium (mmol/L) Platelet count (×109 L–1 ) Serum creatinine (mol/L) GFR (MDRD) (mL/min) Serum albumin (g/L)
Death or LT Alive Death or LT Alive Death or LT Alive Death or LT Alive Death or LT Alive Death or LT Alive Death or LT Alive Death or LT Alive Death or LT Alive
Median (3 months)
n
p
Median (1 year)
n
p
60 [39–83] 56 [24–76] 47 [23–80] 59 [22–100] 1.9 [1.1–3.8] 1.4 [1–3.8] 34 [7–165] 25 [5–487] 136 [119–149] 136 [123–143] 98 [30–310] 136 [35–586] 83 [27–316] 73.5 [35–316] 80.2 [17.7–250.8] 87.35 [19.2–205.8] 30.5 [15–36] 31 [10–43]
17 94 17 94 17 94 17 94 17 94 17 94 17 94 17 94 16 94
0.156
58 [24–83] 55.5 [31–76] 54 [23–84] 63.5 [22–100] 1.5 [1.1–3.8] 1.4 [1–3.8] 25 [7–165] 25.5 [5–487] 136 [119–149] 136 [124–143] 107 [30–456] 140 [35–586] 88 [27–316] 67.5 [21–316] 78.2 [17.7–250.8] 42.9 [19.3–250.8] 30.5 [15–38] 31 [10–43]
43 68 46 68 46 68 46 68 46 68 46 68 46 68 46 68 42 68
0.182
0.003 0.005 0.01 0.733 0.009 0.074 0.113 0.366
0.008 0.065 0.966 0.604 0.105 0.002 0.004 0.354
INR, index normalized ratio; GFR, glomerular filtration rate; MDRD, Modification of Diet in Renal Disease.
Table 2B Univariate analysis of survival according to scores.
Child–Pugh score Modified creatinine Child–Pugh score I Modified creatinine Child–Pugh score II MELD score MELD-Na score Refit MELD score Refit MELD-Na score MESO score iMELD score
Death or LT Alive Death or LT Alive Death or LT Alive Death or LT Alive Death or LT Alive Death or LT Alive Death or LT Alive Death or LT Alive Death or LT Alive
Median (3 months)
n
p
Median (1 year)
n
p
10 [7–13] 9 [7–12] 11 [8–15] 9 [7–15] 11 [6–4] 9 [7–15] 21 [9.4–29.1] 12.4 [6.4–38.8] 22 [9–32] 16.14 [6–39.2] 12.65 [5.5–37.6] 12.65 [5.5–37.6] 12.86 [8.4–24.2] 12.86 [4.8–22.1] 15.11 [6.4–22.5] 9.26 [4.3–29.1] 47.1 [25.2–56.7] 34.8 [22.7–60.8]
17 94 17 94 17 94 17 94 17 94 17 94 17 94 17 94 17 94
0.003
9 [7–13] 9 [7–12] 9 [8–14] 9 [7–15] 9 [8–14] 9 [7–21] 14.56 [7.5–29.1] 12.23 [6.4–38.8] 18.15 [9–32] 15.99 [6–39.2] 15.29 [6.6–25.6] 11.92 [5.5–35.6] 14.21 [8.3–24.2] 12.72 [4.8–22.1] 10.74 [5.3–22.5] 8.99 [4.3–29.1] 37.2 [25.2–56.7] 34.65 [22.7–68.8]
46 68 46 68 46 68 46 68 46 68 46 68 46 68 46 68 46 68
0.175
0.001 <0.001 <0.001 0.005 <0.001 0.052 <0.001 0.001
0.167 0.88 0.023 0.094 0.014 0.036 0.033 0.055
MELD, Model for End Stage Liver Diseases; MESO, MELD to sodium index; TIPS, transjugular portosystemic shunt.
(p < 0.001). If considering LT patients as alive at the date of LT, 1-year actuarial survival rate was 92% for the group A, 70.5% for the group B and 64.2% for the group C (p = 0. 061). Finally, we did an internal validation, repeating the statistical analysis on a group of 50 patients randomly selected with the same mean age and sex
distribution and confirming the prognostic ability of the PLT-MDRD classification (the 1-year actuarial survival rate was 95.1% for the group A, 61.3% for the group B and 31.9% for the group C (p < 0.05)). The only score with significant and independent prognosis ability was the PLT-MDRD classification, with a HR of 6.34
Table 3 Prognostic value of biological variables and scores (expressed as c-statistics). AUROC 3 months MELD Refit MELD Modified creatinine Child–Pugh II MESO Modified creatinine Child–Pugh I iMELD Prothrombin rate Child–Pugh INR MELD-Na Platelet count Bilirubin Refit MELD-Na Creatinine GFR (MDRD)
0.794 0.781 0.775 0.773 0.756 0.752 0.727 0.716 0.716 0.712 0.700 0.696 0.648 0.636 0.621
AUROC 1 year Creatinine GFR (MDRD) Prothrombin rate Refit MELD INR MELD MESO Refit MELD-Na iMELD MELD Na Modified creatinine Child–Pugh II Platelet count Modified creatinine Child–Pugh I Child–Pugh Bilirubin
INR, index normalized ration; MDRD, Modification of Diet in Renal Disease; MELD, Model for End Stage Liver diseases; MESO, MELD to sodium index.
0.677 0.662 0.641 0.640 0.640 0.629 0.620 0.616 0.608 0.595 0.594 0.592 0.577 0.574 0.502
B. Hamel et al. / Digestive and Liver Disease 46 (2014) 1001–1007
1005
Table 4A Actuarial survival rate according to simple clinical and biological parameters.
Age (years) Aetiology Stent Prothrombin rate INR Bilirubin (mol/L) Platelet count (×109 L−1 ) GFR (MDRD)
≥65 <65 Alcohol Others Uncovered Covered ≥50% <50% ≥1.45 <1.45 ≥75 <75 ≥125 <125 ≥90 <90
n
3-month survival (%)
1-year survival (%)
p
24 87 87 24 31 80 80 31 56 55 13 98 62 49 51 60
62.5 87.3 85 70.8 77 83.7 90 60.8 74.9 89.2 53.8 85.9 90.3 71.4 90.2 74.9
45.8 65.2 66.4 41.7 55.1 63.2 64.9 50.7 56.6 65.5 44.9 63.2 68.8 51 70.4 46
0.047
[1.444–27.848] (p = 0.014) for group B and a HR of 12.623 [2.61–61.052] (p = 0.002) for group C. 4. Discussion The present study showed that a simple classification made from platelet count and GFR estimated by MDRD formula was predictive of survival in 111 cirrhotic patients with refractory ascites treated by TIPS. It has been reported that survival after TIPS ranges between 78% and 84% at 3 months, and between 41% and 69% at 1 year. This is in agreement with our results, showing a survival rate without LT of 82% at 3 months and 59.4% at 1 year. Several biological or clinical factors evaluated before TIPS insertion have been identified to have prognostic ability, including age, bilirubin level [8,17–19], prothrombin rate [8,16], platelet count [8], and serum creatinine level [16]. In addition, different scores have been proposed to improve the relevance of simple biological parameters, such as the Child–Pugh score and MELD scores (the latter was initially designed to predict the 3-month mortality after TIPS for refractory ascites or
0.02 0.589 0.059 0.239 0.052 0.024 0.001
the prevention of variceal re-bleeding) [7,16,20–24]. Modifications of these two major scores have been designed to further improve prognostic ability, and were also evaluated in our study: MELDNa, iMELD, MESO, Refit MELD, Refit MELD-Na, modified creatinine Child–Pugh scores I and II. In Supplementary Table S1 the significant scores (and their cutoffs) that significantly predicted survival in previous studies are summarized and compared with our results. Several MELD thresholds have been proposed: 15, 18 (confirmed in our study), and 24. It is clear that a higher MELD score is strongly associated with a higher mortality rate, e.g. the 1-month 40% survival rate for patients with MELD score above 24 from Montgomery et al. [21] and the 57.4% rate from Ferral et al. [23]. Similarly, as Angermayr et al. [16] and Heinzow et al. [24] reported, also our study confirms the poor prognosis after TIPS in Child–Pugh C patients (the 3-month survival rate was 58.9% compared with 91.1% for Child–Pugh B). Recently, Bureau et al. demonstrated that a combination of a bilirubin level below 50 mol/L and a platelet count above 75 × 109 L−1 could represent a simple score to predict survival
Table 4B Actuarial survival rate according to scores. Scores Child–Pugh Modified creatinine Child–Pugh I Modified creatinine Child–Pugh II MELD MELD-Na Refit MELD Refit MELD-Na MESO iMELD PLT < 75 and bilirubin > 50 PLT >75 and bilirubin < 50 PLT-MDRDa
B C ≥10 <10 ≥10 <10 ≥18 <18 ≥17 <17 ≥13 <13 ≥13 <13 ≥10 <10 ≥35.5 <35.5 Yes No Yes No Group A Group B Group C
n
3-month survival (%)
1-year survival (%)
p
79 32 47 64 46 65 28 83 58 53 63 48 64 47 59 52 56 55 8 103 81 30 25 63 23
91.1 58.9 67.9 92.2 69.4 90.8 64.3 88 88.7 75.7 74.5 91.7 78 87.2 72.8 92.3 71.3 92.7 37.5 85.4 87.7 66.5 96 85.7 56.5
67 45.8 54.8 65.6 55.9 64.5 45.4 66.2 69.8 52.8 51.8 72.9 54.1 70.2 50.2 73 53 69.1 37.5 62.8 61.7 59.5 92 58.2 34.8
0.007 0.118 0.194 0.017 0.051 0.019 0.084 0.01 0.048 0.019 0.488 <0.001
MDRD, Modification of Diet in Renal Disease; MELD, Model for End Stage Liver diseases; MESO, MELD to sodium index; INR, index normalized ratio; PLT, platelet count. a PLT-MDRD Group A: PLT > 125 × 109 /L & MDRD > 90 mL/min. PLT-MDRD Group B: PLT > 125 × 109 /L & MDRD < 90 mL/min or PLT < 125 × 109 /l & MDRD > 90 mL/min. PLTMDRD Group C: PLT < 125 × 109 /L & MDRD < 90 mL/min.
1006
B. Hamel et al. / Digestive and Liver Disease 46 (2014) 1001–1007
of patients with refractory ascites treated by TIPS; however, the performance of this simple score was not compared to the MELD or Child–Pugh scores [8]. Despite we do not confirm these results in our cohort, we disclosed that the combination of a bilirubin level above 50 mol/L and a platelet count below 75 × 109 L−1 was associated with poorer survival. Our cohort is largely similar to the different cohorts included in the study of Bureau et al. (one initial cohort and two validation cohorts). Nevertheless, Bureau and colleagues included a greater proportion of Child–Pugh C patients (28.8% in our cohort vs. 46% in their first cohort, 38% in their validation cohort, and 33% in their external cohort). Interestingly, in the intermediate group of the “platelet-bilirubin” classification (bilirubin below 50 mol/L and platelet count below 75 × 109 L−1 or bilirubin above 50 mol/L and platelet count above 75 g/l), the 1year survival rate was 68.2% in our study (similar to patients with bilirubin below 50 mol/L and platelet count above 75 × 109 L−1 , data not shown) and was 31.2% in Bureau’s study (similar to patients with bilirubin above 50 mol/L and platelet count below 75 × 109 L−1 ). This suggests that the patients in the intermediate group included in Bureau’s study were more severe than ours, in agreement with the higher proportion of Child–Pugh C patients. In our study, platelet count and GFR (in the MDRD simplified formula) had a significant and independent prognostic value. The MDRD formula [10] estimates GFR and evaluates renal function more reliably than the creatinine levels, which are directly proportional to muscle mass and disadvantage women and cirrhotic patients. A cut-off of 90 mL/min for the calculated GFR seems rather high, but it must be pointed out that both MDRD and Cockcroft-Gault formulas largely overestimated GFR in patients with decompensated cirrhosis [25]. Therefore, we created a simple classification (“PLT-MDRD”) that was able to clearly distinguish 3 groups of patients according to their 1-year survival. The hazard ratio associated with the intermediate-prognosis group (B) (platelet count above 125 × 109 L−1 and MDRD below 90 mL/min, or platelet count below 125 × 109 L−1 and MDRD above 90 mL/min) was 6.34, and was 12.623 for the poor-prognosis group (C) (platelet count below 125 × 109 L−1 and MDRD below 90 mL/min). In addition, this classification remained the only one statistically significant in multivariate analysis including the Child–Pugh and MELD-derived scores. Compared with the work from Bureau et al., we replaced bilirubin with GFR (in their study, there was no difference in creatinine levels according to survival). It can be hypothesized that bilirubin and prothrombin rate are the best markers for liver function. Platelet count and GFR better reflect the level of portal hypertension and its consequences, especially ascites and renal impairment. In a population of cirrhotic patients with ascites, since patients with severe hepatocellular failure have probably not been considered for TIPS insertion, it is not surprising to find that the “PLT-MDRD” classification has the best prognostic value. A score evaluating portal hypertension would be more discriminating than the assessment of liver function. Indeed, our classification had a better prognostic ability than the MELD score (and its derivatives), despite including creatinine, probably because of the balanced weight of INR and bilirubin. Finally, regarding the efficacy of TIPS in ascites, a remission rate of approximately half of the patients after 1 year has been reported [17,26,27], and few studies have identified significant predictive factors. Taki et al. reported, from a small series of 47 patients, that creatinine and age were predictors of TIPS efficacy [27]. In patients with a creatinine level below 168 mol/l, the remission rate was 76% at 1 year compared with the 50% rate in those with a creatinine level above this cut-off (p = 0.0364). In our work, creatinine was also predictive of efficacy despite not reaching statistical significance (p = 0.05). In conclusion, our study has a number of limitations: first, it is a retrospective study and, secondly, patients were included in the last
13 years, during which technology has developed with the advent of covered stents. Also, the majority of our patients had alcoholic liver disease. Nevertheless, the present remains one of the studies including the largest number of patients (n = 111). We here propose a simple classification, based on platelet count and GFR, to classify cirrhotic patients (mainly alcohol-related) with refractory ascites with regard to their expected 1-year survival. For patients disclosing features of the poor-prognosis group (platelet count below 125 × 109 L−1 and MDRD below 90 mL/min), mortality was approximately of 2/3 at 1 year after TIPS. In such cases, TIPS can probably be contraindicated and other options may be discussed, especially LT. Conflict of interest None declared. 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.dld.2014.06.013. References [1] Gines P, Quintero E, Arroyo V, et al. Compensated cirrhosis: natural history and prognostic factors. Hepatology 1987;7:122–8. [2] Planas R, Balleste B, Alvarez MA, et al. Natural history of decompensated hepatitis C virus-related cirrhosis. A study of 200 patients. Journal of Hepatology 2004;40:823–30. [3] Garcia N, Sanyal AJ. Ascites. Current Treatment Options in Gastroenterology 2001;4:527–37. [4] Zervos EE, Rosemurgy AS. Management of medically refractory ascites. American Journal of Surgery 2001;181:256–64. [5] Salerno F, Camma C, Enea M, et al. Transjugular intrahepatic porto-systemic shunt for refractory ascites: a meta-analysis of individual patient data. Gastroenterology 2007;133:825–34. [6] Wiesner R, Edwards E, Freeman R, et al. Model for end-stage liver disease (MELD) and allocation of donor livers. Gastroenterology 2003;124: 91–6. [7] Malinchoc M, Kamath PS, Gordon FD, et al. A model to predict poor survival in patients undergoing transjugular intrahepatic porto-systemic shunts. Hepatology 2000;31:864–71. [8] Bureau C, Metivier S, D’Amico M, et al. Serum bilirubin and platelet count: a simple predictive model for survival in patients with refractory ascites treated by TIPS. Journal of Hepatology 2011;54:901–7. [9] Moore KP, Wong F, Gines P, et al. The management of ascites in cirrhosis: report on the consensus conference of the International Ascites Club. Hepatology 2003;38:258–66. [10] Levey AS, Bosch JP, Lewis JB, et al. A more accurate method to estimate glomerular filtration rate from serum creatinine: a new prediction equation. Modification of Diet in Renal Disease Study Group. Annals of Internal Medicine 1999;130:461–70. [11] Pugh RN, Murray-Lyon IM, Dawson JL, et al. Transection of the oesophagus for bleeding oesophageal varices. British Journal of Surgery 1973;60: 646–9. [12] Biggins SW, Kim WR, Terrault NA, et al. Evidence-based incorporation of serum sodium concentration into MELD. Gastroenterology 2006;130:1652–60. [13] Luca A, Angermayr B, Bertolini G, et al. An integrated MELD model including serum sodium and age improves the prediction of early mortality in patients with cirrhosis. Liver Transplantation 2007;13:1174–80. [14] Huo TI, Wang YW, Yang YY, et al. Model for end-stage liver disease score to serum sodium ratio index as a prognostic predictor and its correlation with portal pressure in patients with liver cirrhosis. Liver International 2007;27:498–506. [15] Koo JK, Kim JH, Choi YJ, et al. Predictive value of refit model for end-stage liver disease, refit model for end-stage liver disease-Na, and pre-existing scoring system for 3-month mortality in Korean patients with cirrhosis. Journal of Gastroenterology and Hepatology 2013;28:1209–16. [16] Angermayr B, Cejna M, Karnel F, et al. Child–Pugh versus MELD score in predicting survival in patients undergoing transjugular intrahepatic porto-systemic shunt. Gut 2003;52:879–85. [17] Rossle M, Ochs A, Gulberg V, et al. A comparison of paracentesis and transjugular intrahepatic porto-systemic shunting in patients with ascites. New England Journal of Medicine 2000;342:1701–7. [18] Zipprich A, Kuss O, Rogowski S, et al. Incorporating indocyanin green clearance into the Model for End Stage Liver Disease (MELD-ICG) improves prognostic accuracy in intermediate to advanced cirrhosis. Gut 2010;59:963–8. [19] Rajan DK, Haskal ZJ, Clark TW. Serum bilirubin and early mortality after transjugular intrahepatic porto-systemic shunts: results of a multivariate analysis. Journal of Vascular and Interventional Radiology 2002;13:155–61.
B. Hamel et al. / Digestive and Liver Disease 46 (2014) 1001–1007 [20] Alessandria C, Gaia S, Marzano A, et al. Application of the model for end-stage liver disease score for transjugular intrahepatic porto-systemic shunt in cirrhotic patients with refractory ascites and renal impairment. European Journal of Gastroenterology and Hepatology 2004;16:607–12. [21] Montgomery A, Ferral H, Vasan R, et al. MELD score as a predictor of early death in patients undergoing elective transjugular intrahepatic portosystemic shunt (TIPS) procedures. Cardiovascular and Interventional Radiology 2005;28:307–12. [22] Salerno F, Merli M, Cazzaniga M, et al. MELD score is better than Child–Pugh score in predicting 3-month survival of patients undergoing transjugular intrahepatic porto-systemic shunt. Journal of Hepatology 2002;36: 494–500. [23] Ferral H, Gamboa P, Postoak DW, et al. Survival after elective transjugular intrahepatic porto-systemic shunt creation: prediction with model for end-stage liver disease score. Radiology 2004;231:231–6.
1007
[24] Heinzow HS, Lenz P, Kohler M, et al. Clinical outcome and predictors of survival after TIPS insertion in patients with liver cirrhosis. World Journal of Gastroenterology 2012;18:5211–8. [25] Rognant N, Bacchetta J, Dubourg L, et al. What is the best alternative to inulin clearance to estimate GFR in patients with decompensated alcoholic cirrhosis. Nephrology, Dialysis, Transplantation 2010;25:3569–75. [26] Feyssa E, Ortiz J, Grewal K, et al. MELD score less than 15 predicts prolonged survival after transjugular intrahepatic porto-systemic shunt for refractory ascites after liver transplantation. Transplantation 2011;91:786–92. [27] Taki Y, Kanazawa H, Narahara Y, et al. Predictive factors for improvement of ascites after transjugular intrahepatic porto-systemic shunt in patients with refractory ascites. Hepatology Research 2013, http://dx.doi.org/10.1111/hepr.12195 [Epub ahead of print].
Contents lists available at ScienceDirect
Digestive and Liver Disease journal homepage: www.elsevier.com/locate/dld
Liver, Pancreas and Biliary Tract
Prognostic factors in patients with refractory ascites treated by transjugular intrahepatic porto-systemic shunt: From the liver to the kidney Benjamin Hamel a , Olivier Guillaud a , Sabine Roman a,b , Mélanie Vallin a , Frank Pilleul a,b , Pierre-Jean Valette a,b , Luc Henry a , Aymeric Guibal a , Franc¸ois Mion a,b , Jérôme Dumortier a,b,∗ a b
Department of Digestive Diseases, Hospices Civils de Lyon, Edouard Herriot Hospital, France Claude Bernard Lyon 1 University, Lyon, France
a r t i c l e
i n f o
Article history: Received 10 December 2013 Accepted 30 June 2014 Available online 2 August 2014 Keywords: Ascitis Cirrhosis Survival TIPS
a b s t r a c t Background: The aim of this retrospective study was to evaluate the prognostic value of different scores (including Child–Pugh and Model for End Stage Liver Diseases) in cirrhotic patients treated with transjugular intrahepatic porto-systemic shunt for refractory ascites. Methods: Overall, 111 patients with transjugular intrahepatic porto-systemic shunt insertion between January 1998 and July 2012 were included. Results: Survival rates (without transplantation) were 82.0% at 3 months, and 59.4% at 1 year. In addition to standard parameters, a new simple classification based on platelet count and glomerular filtration rate showed strong prognostic ability and could distinguish 3 groups of patients (Log-rank test, p < 0.001): a “good-prognosis” group with platelet counts above 125 × 109 L−1 and a glomerular filtration rate above 90 mL/min (1-year survival rate 92%), a “poor-prognosis” group with platelet counts below 125 × 109 L−1 and a glomerular filtration rate below 90 mL/min (1-year survival rate 34.8%), and an “intermediateprognosis” group (1-year survival rate 58.2%). Multivariate analysis showed a hazard ratio of 6.34 for the intermediate class and of 12.623 for the high class. Conclusions: A new and simple classification including platelet count and glomerular filtration rate is highly predictive of survival in patients with refractory ascites treated with transjugular intrahepatic porto-systemic shunt and could be used to select patients for this procedure. © 2014 Editrice Gastroenterologica Italiana S.r.l. Published by Elsevier Ltd. All rights reserved.
1. Introduction Ascites is the most frequent complication of liver cirrhosis, especially if compared to jaundice, gastrointestinal bleeding, or hepatic encephalopathy [1]. Approximately 50% of cirrhotic patients present with ascites during 10 years of observation [1]. The prognosis for patients with newly established ascites is a 1 and 5-year survival of 85% and 56%, respectively [2]. Life-threatening complications of refractory ascites include spontaneous bacterial peritonitis, abdominal wall hernia (with the risk of incarceration, skin ulceration, or rupture), hepatic hydrothorax, denutrition, and impaired renal function leading to hepatorenal syndrome (HRS).
∗ Corresponding author at: Pavillon Hbis, Hôpital Edouard Herriot, 69437 Lyon Cedex 03, France. Tel.: +33 4 72 11 01 11; fax: +33 4 72 11 01 47. E-mail address: [email protected] (J. Dumortier).
In addition, chronic ascites also strongly affects quality of life [3]. Approximately 90% of cases could be controlled by a combination of optimal diuretic treatment and dietetic rules [1]. The treatment options for diuretic-resistant ascites include therapeutic paracentesis with intravenous albumin infusion, peritoneo-venous shunting, transjugular intrahepatic porto-systemic shunt (TIPS) and liver transplantation (LT) [4]. LT is the only curative treatment, but cannot be largely used because of organ scarcity and extrahepatic contraindications. Several large-scale, randomized, controlled trials of TIPS compared with large-volume paracentesis have demonstrated that TIPS significantly improves the transplant-free survival of cirrhotic patients with refractory ascites [5]. Nevertheless, there is a significant risk of mortality in the first year following TIPS insertion, especially due to worsening of liver function [6]. Therefore, it has been a challenging issue for physicians to elaborate reliable tools for predicting the outcome of patients with refractory ascites and who
http://dx.doi.org/10.1016/j.dld.2014.06.013 1590-8658/© 2014 Editrice Gastroenterologica Italiana S.r.l. Published by Elsevier Ltd. All rights reserved.
1002
B. Hamel et al. / Digestive and Liver Disease 46 (2014) 1001–1007
are candidates for TIPS placement. The Model for End Stage Liver Diseases (MELD) score was initially specifically designed to predict the 3-month mortality risk of patients undergoing elective TIPS [7]. During the past 10 years, in addition to MELD, the prognostic value of several scores has been investigated in this field, including Child–Pugh score, creatinine-modified Child–Pugh score I and II, MELD-Na, iMELD, MESO, and REFIT-MELD. Recently, it has been suggested that a simple score, calculated from bilirubin level and platelet count, could be very efficient for the prediction of survival for patients with refractory ascites treated by TIPS [8]. The aims of the present retrospective study were [1] to evaluate the reliability of different scores (including the standard Child–Pugh and MELD scores and the simple “bilirubin-platelets” score) for the prediction of mortality after TIPS insertion in cirrhotic patients presenting refractory ascites, and [2] to investigate the ability of the same scores to predict ascites control. 2. Patients and methods 2.1. Study population One hundred and eleven consecutive patients who had undergone TIPS placement for refractory ascites complicating liver cirrhosis were included in our centre (Edouard Herriot Hospital, Lyon, France) between January 1998 and July 2012. Refractory ascites was defined as ascites that does not respond to sodium restriction and high-dose diuretic treatment (400 mg/d spironolactone and 160 mg/d furosemide) or that reoccurs rapidly after therapeutic paracentesis or associated with diuretic-induced complications, as proposed by the International Ascites Club [9]. Patients’ demographic characteristics (gender, age) and information on the type of cirrhosis were available for all patients. Baseline biological patients’ characteristics were recorded within one week before TIPS placement and were available for all the patients. Glomerular filtration rate (GFR) was calculated by using the MDRD (Modification of Diet in Renal Disease) simplified formula as follows: 186 × (creatinine/88.4) − 1.154 × age − 0.203 × (0.742 if female gender) × (1.212 if African descent) [10]. 2.2. Calculation of scores The Child–Pugh score was calculated in accordance to Pugh et al. [11]. The other scores were calculated as follows: MELD = 9.6 × loge [creatinine (mg/dL)] + 3.8 × loge [bilirubin (mg/dL)] + 11.2 × loge (INR) + 6.43 [6] [INR: international normalized ratio] MELD-Na = MELD + 1.59 × [135 − Na (mmol/L)]; Na range = 125–140 mmol/L [12] iMELD = MELD + [0.3 × age (years)] − [0.7 × Na (mmol/L) + 100] [13] MESO = [MELD/Na (mmol/L)] × 100 [14] REFIT MELD = 4.082 × log bilirubin (mg/dL) + 8.485 × log creatinine (mg/dL) + 10.671 × log INR + 7.432 [15] REFIT MELD Na = 4.258 × log bilirubin (mg/dL) + 6.792 × log (mg/dL) + 8.290 × log INR + 0.652 log creatinine (140 − Na) − 0.194 × 140 − Na × bilirubin + 6.327 [15] Modified creatinine Child–Pugh score-I (range: 5–19) was derived from the original CP score by adding 0 points for creatinine <1.3 mg/dL and 4 points for creatinine > or = 1.3 mg/dL [16]. Modified creatinine Child–Pugh score-II (range: 5–19) was derived from the original CP score by adding 0 points for
creatinine <1.3 mg/dL, 2 points for creatinine 1.3–1.8 mg/dL and 4 points for creatinine >1.8 mg/dL [15]. 2.3. Patients’ outcome All patients were followed after TIPS insertion. For the analysis of TIPS effectiveness, ascites was considered controlled if patients did not need paracenteses at 6 months. The presence of ascites was assessed and recorded after TIPS revision when this was necessary. For the survival analysis, the follow-up time was defined from the TIPS placement up to death, LT, or last visit (patients who underwent LT were censored as dead). Follow-up was complete through June 30, 2013. 2.4. Statistical analysis Quantitative variables were described using mean, median, range, and standard deviation (SD). Qualitative values were tabulated and percentages were calculated. Quantitative variables were compared using the Student’s t-test or Mann–Whitney test. Qualitative variables were compared using the 2 test. The concordance statistic was calculated both for the scores and tests predicting mortality (at 3 and 12 months), and for those predicting TIPS efficacy (at 6 months). The concordance statistic, c, is identical to the area under the receiver operating characteristic (ROC) curve. The c-statistic varies between 0.5 and 1.0 for sensible models; the higher the value, the better. Only the results of c-statistics above 0.5 were considered as statistically significant. Kaplan–Meier curves were used to represent survival according to the levels of scores and tests. The purpose of the univariate analysis was to select potentially explanatory variables. The continuous variables were dichotomised, by splitting at their median or upper normal value or at previously proposed values, and then compared. A Cox regression model was used to test the significance of these scores and tests as predictors of survival. All tests with a p < 0.1 at univariate analysis were included in the Cox Model. Statistical tests were considered as significant if p < 0.05. All analyses were performed using SPSS (version 18). 3. Results 3.1. Patients’ characteristics The main characteristics of the patients included in the study group are summarized in Table 1. From the whole cohort of 111 patients, 36 subjects were considered to present resistant ascites (no response to high doses of diuretics) and the other 75 presented intractable ascites (with impossibility to give diuretics because of severe side effects or complications). Most of the patients were male, had alcoholic cirrhosis (78.4%), and had a mean age of 56.8 ± 10.7 years. The mean Child–Pugh score was 9 ± 1.45 and mean MELD score was 14 ± 6. No TIPS insertion failure was recorded during the study period for this specific indication. 3.2. TIPS effectiveness After TIPS placement, 24.3% of patients (27/111) underwent a TIPS revision for stenosis and/or thrombosis. The presence of ascites was evaluated and recorded after TIPS revision when this was necessary. The persistence rate of ascites was 45.9% (51/111) at 6 months. None of the biological or clinical parameters considered (including stent type, covered or uncovered) were significantly different when comparing patients with or without ascites persistence. Only
B. Hamel et al. / Digestive and Liver Disease 46 (2014) 1001–1007
1003
Table 1 Characteristics of patients before transjugular intrahepatic portosystemic shunt placement.
Age (years) Male Cause of cirrhosis Alcohol Viral hepatitis Other TIPS Uncovered stent Covered stent Prothrombin index (%) INR Serum bilirubin (mol/L) Serum sodium (mmol/L) Platelet count (×109 L–1 ) Serum creatinine (mol/L) GFR (MDRD) Serum albumin (g/L) (n = 110) Child–Pugh score Child–Pugh class B/C Modified creatinine Child–Pugh score I Modified creatinine Child–Pugh score II MELD score MELD-Na score Refit MELD Refit MELD-Na MESO iMELD
Mean
Median [range]
56.76 ± 11 80 (72.1%)
56 [24–83]
87 (78.4%) 14 (12.6%) 10 (9%) 31 (27.9%) 80 (72.1%) 58.77 ± 16 1.56 ± 0.5 40.84 ± 61
58 [2–100] 1.45 [1–3.8] 25 [5–487]
135.03 ± 5
136 [119–149]
147.41 ± 91
130 [30–586]
89.42 ± 49
74 [27–316]
94.2 ± 43.2 30.47 ± 6
86.1 [17.7–250.8] 31 [10–43]
9±1 79 (71.2%)/32 (28.8%) 10 ± 2
9 [7–13]
9±2
9 [7–15]
14 ± 6 17 ± 6 14 ± 6 13 ± 6 11 ± 4 37 ± 7
13.1 [6.4–38.8] 16.8 [6–39.2] 13.24 [5.5–37.6] 12.96 [4.8–35.2] 9.77 [4.3–29.1] 35.5 [22.7–60.8]
9 [7–15]
INR, index normalized ratio; MDRD, Modification of Diet in Renal Disease; MELD, Model for End Stage Liver diseases; MESO, MELD to sodium index.
the difference in the initial creatinine level was close to statistical significance: 98.5 ± 58 mol/L for patients with ascites vs. 81.7 ± 8 mol/L for patients without ascites (p = 0.05). 3.3. Survival After TIPS insertion, overall actuarial survival rates (without LT) were 82% at 3 months, 72.9% at 6 months, 59.4% at 1 year, and 38.5% at 5 years: 19 patients died or underwent LT at 3 months (4 LT and 15 deaths), and 43 patients died or underwent LT at 1 year (16 LT and 27 deaths). 3.4. Prognostic factors Table 2 lists the results of univariate survival analysis according to clinical and biological variables (Table 2A) and scores (Table 2B). All the following parameters were associated with a significantly better 3-month survival: higher prothrombin level, lower INR, lower Child–Pugh score, lower modified creatinine Child Pugh scores I and II, lower MELD score, lower MELD-Na score, lower Refit MELD, lower MESO score, and lower iMELD score. Similarly, higher TP level, higher MDRD, lower creatinine, lower MELD score, lower Refit MELD and Refit MELD-Na, and lower iMELD were all significantly associated with a better 1-year survival. When survival was analyzed according to gender, 83.8% of men and 87.1% of women were found to be alive at 3 months (p = 0.66), and 61.3% of men and 61.3% of women were alive at 1 year (p = 0.997). Among patients with alcohol-related cirrhosis, 75/87 (86.2%) were alive at 3 months compared with 19/24 (79.2%) of those with non-alcoholic cirrhosis (p = 0.397). At 1 year, 58/87
Fig. 1. Actuarial rates of survival according to platelet-MDRD classification (Kaplan–Meier curves, log-rank p value < 0.001).
(66.7%) of patients with alcoholic cirrhosis and 10/24 (41.7%) of those with non-alcoholic cirrhosis were alive (p = 0.026). Table 3 shows the prognostic ability (expressed as c-statistics) of the biological parameters and scores for the two periods of followup (3 months and 1 year). While MELD and Refit MELD scores had the highest prognostic ability for 3-month survival, creatinine and GFR had the highest prognostic ability for 1-year survival. The ability of both clinical and simple biological parameters to predict actuarial 1-year survival was evaluated according to the Log-rank test (Table 4A). Platelet count, alcoholic origin of cirrhosis, and MDRD were associated with a significantly better survival. Multivariate analysis disclosed that variables with a significant prognostic ability were a platelet count below 125 × 109 L−1 [HR 2.096 (1.129–3.891), p = 0.019], an MDRD above 90 mL/min [HR 3.234 (1.611–6.495), p = 0.001], and the alcoholic origin of cirrhosis [HR 0.506 (0.265–0.966), p = 0.039]. The ability of the scores to predict actuarial 1-year survival was evaluated with the Log-rank test (Table 4B). The following parameters were all associated with a significantly worse survival: Child–Pugh class C, MELD score above 18, MELD score above 13, Refit MELD score above 13, MESO score greater than 10, and iMELD score above 35.5. The combination of a bilirubin level below 50 mol/L and a platelet count above 75 × 109 L−1 was not predictive of survival. Nevertheless, the combination of a bilirubin level above 50 mol/L and a platelet count below 75 × 109 L−1 was predictive of survival: the 1-year survival rate of patients with platelet count above 75 × 109 L−1 or a bilirubin level lower than 50 mol/L was 62.8%, and was 37.5% for patients with platelet count below 75 × 109 L−1 and bilirubin level above 50 mol/L (p = 0.019). Since prognosis was significantly and independently modified in our cohort according to platelet count and MDRD estimation of GFR (from multivariate analysis), we developed a new and simple classification, the PLT-MDRD. This classification divided patients into three groups (Fig. 1): the “good-prognosis” group (group A, n = 25), with a platelet count above 125 × 109 L−1 and an MDRD over 90 mL/min, the “intermediate-prognosis” group (group B, n = 63), with a platelet count above 125 × 109 L−1 and an MDRD below 90 mL/min or platelet count below 125 × 109 L−1 and an MDRD above 90 mL/min, and the “poor-prognosis” group (group C, n = 23), with a platelet count below 125 × 109 L−1 and an MDRD below 90 mL/min. The 1-year actuarial survival rate was 92% for group A, 58.2% for group B, and 34.8% for group C
1004
B. Hamel et al. / Digestive and Liver Disease 46 (2014) 1001–1007
Table 2A Univariate analysis of survival according to clinical and biological data.
Age (years) Prothrombin rate INR Serum bilirubin (mol/L) Serum sodium (mmol/L) Platelet count (×109 L–1 ) Serum creatinine (mol/L) GFR (MDRD) (mL/min) Serum albumin (g/L)
Death or LT Alive Death or LT Alive Death or LT Alive Death or LT Alive Death or LT Alive Death or LT Alive Death or LT Alive Death or LT Alive Death or LT Alive
Median (3 months)
n
p
Median (1 year)
n
p
60 [39–83] 56 [24–76] 47 [23–80] 59 [22–100] 1.9 [1.1–3.8] 1.4 [1–3.8] 34 [7–165] 25 [5–487] 136 [119–149] 136 [123–143] 98 [30–310] 136 [35–586] 83 [27–316] 73.5 [35–316] 80.2 [17.7–250.8] 87.35 [19.2–205.8] 30.5 [15–36] 31 [10–43]
17 94 17 94 17 94 17 94 17 94 17 94 17 94 17 94 16 94
0.156
58 [24–83] 55.5 [31–76] 54 [23–84] 63.5 [22–100] 1.5 [1.1–3.8] 1.4 [1–3.8] 25 [7–165] 25.5 [5–487] 136 [119–149] 136 [124–143] 107 [30–456] 140 [35–586] 88 [27–316] 67.5 [21–316] 78.2 [17.7–250.8] 42.9 [19.3–250.8] 30.5 [15–38] 31 [10–43]
43 68 46 68 46 68 46 68 46 68 46 68 46 68 46 68 42 68
0.182
0.003 0.005 0.01 0.733 0.009 0.074 0.113 0.366
0.008 0.065 0.966 0.604 0.105 0.002 0.004 0.354
INR, index normalized ratio; GFR, glomerular filtration rate; MDRD, Modification of Diet in Renal Disease.
Table 2B Univariate analysis of survival according to scores.
Child–Pugh score Modified creatinine Child–Pugh score I Modified creatinine Child–Pugh score II MELD score MELD-Na score Refit MELD score Refit MELD-Na score MESO score iMELD score
Death or LT Alive Death or LT Alive Death or LT Alive Death or LT Alive Death or LT Alive Death or LT Alive Death or LT Alive Death or LT Alive Death or LT Alive
Median (3 months)
n
p
Median (1 year)
n
p
10 [7–13] 9 [7–12] 11 [8–15] 9 [7–15] 11 [6–4] 9 [7–15] 21 [9.4–29.1] 12.4 [6.4–38.8] 22 [9–32] 16.14 [6–39.2] 12.65 [5.5–37.6] 12.65 [5.5–37.6] 12.86 [8.4–24.2] 12.86 [4.8–22.1] 15.11 [6.4–22.5] 9.26 [4.3–29.1] 47.1 [25.2–56.7] 34.8 [22.7–60.8]
17 94 17 94 17 94 17 94 17 94 17 94 17 94 17 94 17 94
0.003
9 [7–13] 9 [7–12] 9 [8–14] 9 [7–15] 9 [8–14] 9 [7–21] 14.56 [7.5–29.1] 12.23 [6.4–38.8] 18.15 [9–32] 15.99 [6–39.2] 15.29 [6.6–25.6] 11.92 [5.5–35.6] 14.21 [8.3–24.2] 12.72 [4.8–22.1] 10.74 [5.3–22.5] 8.99 [4.3–29.1] 37.2 [25.2–56.7] 34.65 [22.7–68.8]
46 68 46 68 46 68 46 68 46 68 46 68 46 68 46 68 46 68
0.175
0.001 <0.001 <0.001 0.005 <0.001 0.052 <0.001 0.001
0.167 0.88 0.023 0.094 0.014 0.036 0.033 0.055
MELD, Model for End Stage Liver Diseases; MESO, MELD to sodium index; TIPS, transjugular portosystemic shunt.
(p < 0.001). If considering LT patients as alive at the date of LT, 1-year actuarial survival rate was 92% for the group A, 70.5% for the group B and 64.2% for the group C (p = 0. 061). Finally, we did an internal validation, repeating the statistical analysis on a group of 50 patients randomly selected with the same mean age and sex
distribution and confirming the prognostic ability of the PLT-MDRD classification (the 1-year actuarial survival rate was 95.1% for the group A, 61.3% for the group B and 31.9% for the group C (p < 0.05)). The only score with significant and independent prognosis ability was the PLT-MDRD classification, with a HR of 6.34
Table 3 Prognostic value of biological variables and scores (expressed as c-statistics). AUROC 3 months MELD Refit MELD Modified creatinine Child–Pugh II MESO Modified creatinine Child–Pugh I iMELD Prothrombin rate Child–Pugh INR MELD-Na Platelet count Bilirubin Refit MELD-Na Creatinine GFR (MDRD)
0.794 0.781 0.775 0.773 0.756 0.752 0.727 0.716 0.716 0.712 0.700 0.696 0.648 0.636 0.621
AUROC 1 year Creatinine GFR (MDRD) Prothrombin rate Refit MELD INR MELD MESO Refit MELD-Na iMELD MELD Na Modified creatinine Child–Pugh II Platelet count Modified creatinine Child–Pugh I Child–Pugh Bilirubin
INR, index normalized ration; MDRD, Modification of Diet in Renal Disease; MELD, Model for End Stage Liver diseases; MESO, MELD to sodium index.
0.677 0.662 0.641 0.640 0.640 0.629 0.620 0.616 0.608 0.595 0.594 0.592 0.577 0.574 0.502
B. Hamel et al. / Digestive and Liver Disease 46 (2014) 1001–1007
1005
Table 4A Actuarial survival rate according to simple clinical and biological parameters.
Age (years) Aetiology Stent Prothrombin rate INR Bilirubin (mol/L) Platelet count (×109 L−1 ) GFR (MDRD)
≥65 <65 Alcohol Others Uncovered Covered ≥50% <50% ≥1.45 <1.45 ≥75 <75 ≥125 <125 ≥90 <90
n
3-month survival (%)
1-year survival (%)
p
24 87 87 24 31 80 80 31 56 55 13 98 62 49 51 60
62.5 87.3 85 70.8 77 83.7 90 60.8 74.9 89.2 53.8 85.9 90.3 71.4 90.2 74.9
45.8 65.2 66.4 41.7 55.1 63.2 64.9 50.7 56.6 65.5 44.9 63.2 68.8 51 70.4 46
0.047
[1.444–27.848] (p = 0.014) for group B and a HR of 12.623 [2.61–61.052] (p = 0.002) for group C. 4. Discussion The present study showed that a simple classification made from platelet count and GFR estimated by MDRD formula was predictive of survival in 111 cirrhotic patients with refractory ascites treated by TIPS. It has been reported that survival after TIPS ranges between 78% and 84% at 3 months, and between 41% and 69% at 1 year. This is in agreement with our results, showing a survival rate without LT of 82% at 3 months and 59.4% at 1 year. Several biological or clinical factors evaluated before TIPS insertion have been identified to have prognostic ability, including age, bilirubin level [8,17–19], prothrombin rate [8,16], platelet count [8], and serum creatinine level [16]. In addition, different scores have been proposed to improve the relevance of simple biological parameters, such as the Child–Pugh score and MELD scores (the latter was initially designed to predict the 3-month mortality after TIPS for refractory ascites or
0.02 0.589 0.059 0.239 0.052 0.024 0.001
the prevention of variceal re-bleeding) [7,16,20–24]. Modifications of these two major scores have been designed to further improve prognostic ability, and were also evaluated in our study: MELDNa, iMELD, MESO, Refit MELD, Refit MELD-Na, modified creatinine Child–Pugh scores I and II. In Supplementary Table S1 the significant scores (and their cutoffs) that significantly predicted survival in previous studies are summarized and compared with our results. Several MELD thresholds have been proposed: 15, 18 (confirmed in our study), and 24. It is clear that a higher MELD score is strongly associated with a higher mortality rate, e.g. the 1-month 40% survival rate for patients with MELD score above 24 from Montgomery et al. [21] and the 57.4% rate from Ferral et al. [23]. Similarly, as Angermayr et al. [16] and Heinzow et al. [24] reported, also our study confirms the poor prognosis after TIPS in Child–Pugh C patients (the 3-month survival rate was 58.9% compared with 91.1% for Child–Pugh B). Recently, Bureau et al. demonstrated that a combination of a bilirubin level below 50 mol/L and a platelet count above 75 × 109 L−1 could represent a simple score to predict survival
Table 4B Actuarial survival rate according to scores. Scores Child–Pugh Modified creatinine Child–Pugh I Modified creatinine Child–Pugh II MELD MELD-Na Refit MELD Refit MELD-Na MESO iMELD PLT < 75 and bilirubin > 50 PLT >75 and bilirubin < 50 PLT-MDRDa
B C ≥10 <10 ≥10 <10 ≥18 <18 ≥17 <17 ≥13 <13 ≥13 <13 ≥10 <10 ≥35.5 <35.5 Yes No Yes No Group A Group B Group C
n
3-month survival (%)
1-year survival (%)
p
79 32 47 64 46 65 28 83 58 53 63 48 64 47 59 52 56 55 8 103 81 30 25 63 23
91.1 58.9 67.9 92.2 69.4 90.8 64.3 88 88.7 75.7 74.5 91.7 78 87.2 72.8 92.3 71.3 92.7 37.5 85.4 87.7 66.5 96 85.7 56.5
67 45.8 54.8 65.6 55.9 64.5 45.4 66.2 69.8 52.8 51.8 72.9 54.1 70.2 50.2 73 53 69.1 37.5 62.8 61.7 59.5 92 58.2 34.8
0.007 0.118 0.194 0.017 0.051 0.019 0.084 0.01 0.048 0.019 0.488 <0.001
MDRD, Modification of Diet in Renal Disease; MELD, Model for End Stage Liver diseases; MESO, MELD to sodium index; INR, index normalized ratio; PLT, platelet count. a PLT-MDRD Group A: PLT > 125 × 109 /L & MDRD > 90 mL/min. PLT-MDRD Group B: PLT > 125 × 109 /L & MDRD < 90 mL/min or PLT < 125 × 109 /l & MDRD > 90 mL/min. PLTMDRD Group C: PLT < 125 × 109 /L & MDRD < 90 mL/min.
1006
B. Hamel et al. / Digestive and Liver Disease 46 (2014) 1001–1007
of patients with refractory ascites treated by TIPS; however, the performance of this simple score was not compared to the MELD or Child–Pugh scores [8]. Despite we do not confirm these results in our cohort, we disclosed that the combination of a bilirubin level above 50 mol/L and a platelet count below 75 × 109 L−1 was associated with poorer survival. Our cohort is largely similar to the different cohorts included in the study of Bureau et al. (one initial cohort and two validation cohorts). Nevertheless, Bureau and colleagues included a greater proportion of Child–Pugh C patients (28.8% in our cohort vs. 46% in their first cohort, 38% in their validation cohort, and 33% in their external cohort). Interestingly, in the intermediate group of the “platelet-bilirubin” classification (bilirubin below 50 mol/L and platelet count below 75 × 109 L−1 or bilirubin above 50 mol/L and platelet count above 75 g/l), the 1year survival rate was 68.2% in our study (similar to patients with bilirubin below 50 mol/L and platelet count above 75 × 109 L−1 , data not shown) and was 31.2% in Bureau’s study (similar to patients with bilirubin above 50 mol/L and platelet count below 75 × 109 L−1 ). This suggests that the patients in the intermediate group included in Bureau’s study were more severe than ours, in agreement with the higher proportion of Child–Pugh C patients. In our study, platelet count and GFR (in the MDRD simplified formula) had a significant and independent prognostic value. The MDRD formula [10] estimates GFR and evaluates renal function more reliably than the creatinine levels, which are directly proportional to muscle mass and disadvantage women and cirrhotic patients. A cut-off of 90 mL/min for the calculated GFR seems rather high, but it must be pointed out that both MDRD and Cockcroft-Gault formulas largely overestimated GFR in patients with decompensated cirrhosis [25]. Therefore, we created a simple classification (“PLT-MDRD”) that was able to clearly distinguish 3 groups of patients according to their 1-year survival. The hazard ratio associated with the intermediate-prognosis group (B) (platelet count above 125 × 109 L−1 and MDRD below 90 mL/min, or platelet count below 125 × 109 L−1 and MDRD above 90 mL/min) was 6.34, and was 12.623 for the poor-prognosis group (C) (platelet count below 125 × 109 L−1 and MDRD below 90 mL/min). In addition, this classification remained the only one statistically significant in multivariate analysis including the Child–Pugh and MELD-derived scores. Compared with the work from Bureau et al., we replaced bilirubin with GFR (in their study, there was no difference in creatinine levels according to survival). It can be hypothesized that bilirubin and prothrombin rate are the best markers for liver function. Platelet count and GFR better reflect the level of portal hypertension and its consequences, especially ascites and renal impairment. In a population of cirrhotic patients with ascites, since patients with severe hepatocellular failure have probably not been considered for TIPS insertion, it is not surprising to find that the “PLT-MDRD” classification has the best prognostic value. A score evaluating portal hypertension would be more discriminating than the assessment of liver function. Indeed, our classification had a better prognostic ability than the MELD score (and its derivatives), despite including creatinine, probably because of the balanced weight of INR and bilirubin. Finally, regarding the efficacy of TIPS in ascites, a remission rate of approximately half of the patients after 1 year has been reported [17,26,27], and few studies have identified significant predictive factors. Taki et al. reported, from a small series of 47 patients, that creatinine and age were predictors of TIPS efficacy [27]. In patients with a creatinine level below 168 mol/l, the remission rate was 76% at 1 year compared with the 50% rate in those with a creatinine level above this cut-off (p = 0.0364). In our work, creatinine was also predictive of efficacy despite not reaching statistical significance (p = 0.05). In conclusion, our study has a number of limitations: first, it is a retrospective study and, secondly, patients were included in the last
13 years, during which technology has developed with the advent of covered stents. Also, the majority of our patients had alcoholic liver disease. Nevertheless, the present remains one of the studies including the largest number of patients (n = 111). We here propose a simple classification, based on platelet count and GFR, to classify cirrhotic patients (mainly alcohol-related) with refractory ascites with regard to their expected 1-year survival. For patients disclosing features of the poor-prognosis group (platelet count below 125 × 109 L−1 and MDRD below 90 mL/min), mortality was approximately of 2/3 at 1 year after TIPS. In such cases, TIPS can probably be contraindicated and other options may be discussed, especially LT. Conflict of interest None declared. 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.dld.2014.06.013. References [1] Gines P, Quintero E, Arroyo V, et al. Compensated cirrhosis: natural history and prognostic factors. Hepatology 1987;7:122–8. [2] Planas R, Balleste B, Alvarez MA, et al. Natural history of decompensated hepatitis C virus-related cirrhosis. A study of 200 patients. Journal of Hepatology 2004;40:823–30. [3] Garcia N, Sanyal AJ. Ascites. Current Treatment Options in Gastroenterology 2001;4:527–37. [4] Zervos EE, Rosemurgy AS. Management of medically refractory ascites. American Journal of Surgery 2001;181:256–64. [5] Salerno F, Camma C, Enea M, et al. Transjugular intrahepatic porto-systemic shunt for refractory ascites: a meta-analysis of individual patient data. Gastroenterology 2007;133:825–34. [6] Wiesner R, Edwards E, Freeman R, et al. Model for end-stage liver disease (MELD) and allocation of donor livers. Gastroenterology 2003;124: 91–6. [7] Malinchoc M, Kamath PS, Gordon FD, et al. A model to predict poor survival in patients undergoing transjugular intrahepatic porto-systemic shunts. Hepatology 2000;31:864–71. [8] Bureau C, Metivier S, D’Amico M, et al. Serum bilirubin and platelet count: a simple predictive model for survival in patients with refractory ascites treated by TIPS. Journal of Hepatology 2011;54:901–7. [9] Moore KP, Wong F, Gines P, et al. The management of ascites in cirrhosis: report on the consensus conference of the International Ascites Club. Hepatology 2003;38:258–66. [10] Levey AS, Bosch JP, Lewis JB, et al. A more accurate method to estimate glomerular filtration rate from serum creatinine: a new prediction equation. Modification of Diet in Renal Disease Study Group. Annals of Internal Medicine 1999;130:461–70. [11] Pugh RN, Murray-Lyon IM, Dawson JL, et al. Transection of the oesophagus for bleeding oesophageal varices. British Journal of Surgery 1973;60: 646–9. [12] Biggins SW, Kim WR, Terrault NA, et al. Evidence-based incorporation of serum sodium concentration into MELD. Gastroenterology 2006;130:1652–60. [13] Luca A, Angermayr B, Bertolini G, et al. An integrated MELD model including serum sodium and age improves the prediction of early mortality in patients with cirrhosis. Liver Transplantation 2007;13:1174–80. [14] Huo TI, Wang YW, Yang YY, et al. Model for end-stage liver disease score to serum sodium ratio index as a prognostic predictor and its correlation with portal pressure in patients with liver cirrhosis. Liver International 2007;27:498–506. [15] Koo JK, Kim JH, Choi YJ, et al. Predictive value of refit model for end-stage liver disease, refit model for end-stage liver disease-Na, and pre-existing scoring system for 3-month mortality in Korean patients with cirrhosis. Journal of Gastroenterology and Hepatology 2013;28:1209–16. [16] Angermayr B, Cejna M, Karnel F, et al. Child–Pugh versus MELD score in predicting survival in patients undergoing transjugular intrahepatic porto-systemic shunt. Gut 2003;52:879–85. [17] Rossle M, Ochs A, Gulberg V, et al. A comparison of paracentesis and transjugular intrahepatic porto-systemic shunting in patients with ascites. New England Journal of Medicine 2000;342:1701–7. [18] Zipprich A, Kuss O, Rogowski S, et al. Incorporating indocyanin green clearance into the Model for End Stage Liver Disease (MELD-ICG) improves prognostic accuracy in intermediate to advanced cirrhosis. Gut 2010;59:963–8. [19] Rajan DK, Haskal ZJ, Clark TW. Serum bilirubin and early mortality after transjugular intrahepatic porto-systemic shunts: results of a multivariate analysis. Journal of Vascular and Interventional Radiology 2002;13:155–61.
B. Hamel et al. / Digestive and Liver Disease 46 (2014) 1001–1007 [20] Alessandria C, Gaia S, Marzano A, et al. Application of the model for end-stage liver disease score for transjugular intrahepatic porto-systemic shunt in cirrhotic patients with refractory ascites and renal impairment. European Journal of Gastroenterology and Hepatology 2004;16:607–12. [21] Montgomery A, Ferral H, Vasan R, et al. MELD score as a predictor of early death in patients undergoing elective transjugular intrahepatic portosystemic shunt (TIPS) procedures. Cardiovascular and Interventional Radiology 2005;28:307–12. [22] Salerno F, Merli M, Cazzaniga M, et al. MELD score is better than Child–Pugh score in predicting 3-month survival of patients undergoing transjugular intrahepatic porto-systemic shunt. Journal of Hepatology 2002;36: 494–500. [23] Ferral H, Gamboa P, Postoak DW, et al. Survival after elective transjugular intrahepatic porto-systemic shunt creation: prediction with model for end-stage liver disease score. Radiology 2004;231:231–6.
1007
[24] Heinzow HS, Lenz P, Kohler M, et al. Clinical outcome and predictors of survival after TIPS insertion in patients with liver cirrhosis. World Journal of Gastroenterology 2012;18:5211–8. [25] Rognant N, Bacchetta J, Dubourg L, et al. What is the best alternative to inulin clearance to estimate GFR in patients with decompensated alcoholic cirrhosis. Nephrology, Dialysis, Transplantation 2010;25:3569–75. [26] Feyssa E, Ortiz J, Grewal K, et al. MELD score less than 15 predicts prolonged survival after transjugular intrahepatic porto-systemic shunt for refractory ascites after liver transplantation. Transplantation 2011;91:786–92. [27] Taki Y, Kanazawa H, Narahara Y, et al. Predictive factors for improvement of ascites after transjugular intrahepatic porto-systemic shunt in patients with refractory ascites. Hepatology Research 2013, http://dx.doi.org/10.1111/hepr.12195 [Epub ahead of print].