Child-Pugh Scores And MELD Scores Predict The Outcomes Of ERCP In Cirrhotic Patients With Choledocholithiasis

Background and Objection:The incidence of gallstones has been found to be more common in cirrhotic patients than that in the non-cirrhotic population. Furthermore, different studies have showed that the frequency of common bile stones in cirrhotic patients is three times higher than that in non-cirrhotic patients. On the other hand, several reports have confirmed that the rate of acute cholangitis and acute obstructive suppurative cholangitis (AOSC), which was associated with high risk of death, were 44% and 19% in cirrhotic patients with choledocholithiasis, respectively. As is well-known, endoscopic retrograde cholangiopancreatography (ERCP) has become the first-line treatment for common bile stones in the world. However, the main complications, such as post-ERCP pancreatitis (PEP) or hemorrhage, associated with this operation are unavoidable even in general population. These complications may be associated with endoscopic papillary balloon dilatation (EPBD) and/or endoscopic sphincterotomy (EST) for clearing the stones. But, gastro-esophageal varices, coagulation dysfunction, liver function disorder and ascites are very common in cirrhotic patients, and these abnormal conditions may increase the risk of complications. Due to the intra-operative or/and postoperative hemorrhage, the rates of mortality have been proven to be significantly higher in the cirrhotic patients than that in control group.Although the complications and mortality rates of ERCP are higher in cirrhotic patients than that in general population. But, compared with the outcomes of surgery and laparoscopy for choledocholithiasis in cirrhotic patients, ERCP has lower complications and mortality rates. So it has been recommended as the initial choice of management for choledocholithiasis. However, those reports, which showed the effect and safety of ERCP, included only a small number of cases. Therefore, more studies or larger sample sizes are needed to determine a clear conclusion. In addition, no study was focused on which indicator could predict the morbidity and mortality related to this procedure.The two scores, the Child-Pugh scores and the Model for end stage liver disease (MELD) scores, are the main indicators of assessment for the condition in cirrhotic patients. The Child-Pugh scores, which are based on serum albumin, serum bilirubin, prothrombin time (PT), ascites and encephalopathy, has been widely used to predict the risk of death and operative complications in patients with cirrhosis. Because of its simplicity as it can be calculated at the bed side, the Child-Pugh scores obtain the popularity among the medical workers. MELD scores, which are only based on three simple and objective laboratory variables (serum bilirubin, serum creatinine and prothrombin time(PT)) and the etiology of cirrhosis, were initially used to predict the mortality after transjugular intrahepatic portosystemic shunt. And some reports suggested that Child-Pugh scores have lower predictive value according to cirrhotic patients than MELD scores because of two subjective parameters (ascites and encephalopathy). Furthermore, MELD-Na scores have more predictive value than MELD in some reports because the serum sodium was an independent risk factor of mortality in cirrhotic patients. But, no study has been reported about the predictive value of those scores in cirrhotic patients with choledocholithiasis after ERCP operation by now. We evaluated the safety and efficacy of ERCP in cirrhotic patients with choledocholithiasis and accessed the Child-Pugh scores, MELD scores and MELD-Na scores for prediction of morbidity and mortality.Methods and materialsIt was a retrospective study in a single endoscopic center. Between January 2000 and June 2014,7829 ERCP operations were performed in Taizhou Hospital of Zhejiang Province. Eighty cirrhotic patients with choledocholithiasis underwent ERCP operations (118 times), including one time of ERCP in 59 patients and two or more times in 21 patients. Except 3 unsuccessful cases, the data of 77 consecutive cirrhotic patients for the first time of ERCP were collected in this study. The diagnosis was based on the clinical history, physical examination, laboratory tests, outcomes of abdominal ultrasonography and/or CT scan, and endoscopic findings. Child-Pugh scores were calculated using preoperative PT, serum albumin, serum total bilirubin, ascites and encephalopathy for every patient. And Child-Pugh classifications were defined as class A (5-6), class B (7-9), and class C (10-15). MELD scores were based on the serum creatinine (Cr), serum total bilirubin (TBil), international normalized ratio (INR) and etiology. The following formula was used for calculating the MELD scores:MELD=9.57*loge (Cr mg/dL)+3.78* loge (TBil mg/dL)+11.2* loge (INR)+6.43*(etiology). The following formula was used for calculating the MELD-Na scores:MELD-Na=MELD scores+1.59*(135-serum sodium (mmol/L)).Fresh-frozen plasma was administered in patients with coagulopathy (INR>1.8), and platelet transfusions were performed in patients with platelet counts less than 30,000/mm3. In case of patients with INR>1.8 or PT> 18, vitamin K1(30 mg) was routinely administered by intramuscular injection. For patients with Child-Pugh classifications A or B, pethidine (50 mg), scopolamine (40 mg) and diazepam (5 mg) were conventionally administered by intramuscular injection for intraoperative sedation. In order to prevent hepatic encephalopathy, only scopolamine and pethidine at the same dose were administered for patients with classification C. EST was performed if the size of the stone was>1 cm and<2 cm, and EPBD for≤1cm. If the size of the stone were≥2 cm, presence of multiple stones, or Child-Pugh classification C, a plastic biliary stent or endoscopic nasobiliary drainage (ENBD) was performed. ENBD was performed for patients with AOSC. Somatostatin was administered if the patient with gastric and/or esophageal varices. Other intraoperative and postoperative managements were the stand ERCP operations. One author got the patients’survival situation by telephone follow-up interviews on July 11,2014 and July 15,2014.Statistical analysis was performed using IBM SPSS software, version 19.0 (IBM, Armonk, NY, USA). Associations involving parametric data were assessed using Student’s t test. Dichotomous nonparametric data were assessed using the χ2 test or Fisher’s exact test. Coordinate points of a receiver-operating characteristic (ROC) curve were used to identify significant Child-Pugh scores, MELD scores or MELD-Na scores points, including the influence of subgroups. The area under curve (AUC)> 0.7 was considered middle or high predictive value for deciding the critical value. The Kaplan-Meier method with a log-rank test comparison was used for survival analysis. P< 0.05 was considered statistically significant.Results:In total,77 cirrhotic patients with choledocholithiasis were enrolled in the study, and the mean ages were 62.6±11.2 years with sex ratio 41:36 (male:female). Thirteen patients had a preoperative cholecystectomy because of gallbladder stones, and the rest (1/14) had a procedure for gallbladder polyps. The total incidence rate of gallbladder stones was 54.5%(42/77) in cirrhotic patients with choledocholithiasis. The rate of acute pancreatitis, AOSC, infectious shock, and hepatic encephalopathy before operation was 14.3%,6.5%,5.2% and 0%, respectively. And biliary colic (83.1%), jaundice (71.4%), fever (26.0%) and ascites (30.0%) were the main symptoms. Hepatitis B (54.5%) was the main etiology of cirrhosis, and the others were secondary biliary (33.8%), alcohol (3.9%), schistosome (2.6%), cryptogenic (1.3%), hepatitis C (1.3%) and mixed etiology (2.6%).Based on the operation of ERCP, the rate of EST, EPBD, single ENBD and biliary stents was 51.9%,32.5%,10.4% and 5.2%, respectively. A total of 31 complications, occurred in 24 patients, were successfully dealt with endoscopy or/and medicine without surgical treatment or death. The rate of intraoperative hemorrhage, postoperative hemorrhage, PEP, aggravated infection of biliary tract, hepatic encephalopathy and respiratory failure was 13.0%,18.2%,6.1%,1.4%,1.3% and 1.3%, respectively, including 4 patients with both intra- and postoperative hemorrhage. The rate of intraoperative hemorrhage had no statistically significant difference was found between EST with no-EST (12.5% vs.13.5%,χ2=0,P=1), and the same outcomes according to postoperative hemorrhage (22.5% vs.13.5%, χ2 =1.043, P=0.307). However, the rate of PEP had statistically significant difference between the two groups (16% vs.0%, χ2=3.202, P=0.074). Eleven (14.3%)patients needed kinds of transfusion, including red blood cells (6.5%), plasma & platelet (3.9%), plasma & red blood cells (2.6%) and red blood cells & platelet (1.3%).The average scores of Child-Pugh scores, MELD scores and MELD-Na were 8.1±2.2 (range from 5 to 13),13.7±5.5 (range from 5 to 26) and 16.4±7.5 (range from 5 to 35.1), respectively. The average Child-Pugh scores had also statistically significant higher in patients with complications than in those without complication (9.1±1.8 vs.7.6±2.2, t=2.951, P=0.004). However, ROC analysis identified Child-Pugh scores had no value for predicting the incidence of complications (AUC= 0.69,95% confidence interval (CI)=57.4%-81.1%; P=0.007). And no significant difference in the rate of complications was observed among patients with different Child-Pugh classifications. The average MELD scores had statistically significant higher in patients with complications than in those without complication (17.1±5.4 vs. 12.2±4.9, t=3.911, P=0.000). ROC analysis identified MELD scores higher than 11.5 as the best cutoff value for predicting the incidence of complications (AUC= 0.75,95% confidence interval (CI)=63.0%-86.5%; sensitivity=87.5%; specificity =50.9%, P=0.000). In total,21 (44.7%) cases developed a complication among patients with MELD scores higher 11.5, and 3 (10%) cases with a lower MELD score developed a complication (χ2=10.266, P=0.001). PEP, intraoperative hemorrhage, and postoperative hemorrhaging occurred in two patients with MELD scores of 23 and 21, respectively. PEP and postoperative hemorrhaging occurred in two patients with MELD scores of 24 and 14, respectively. Intra- and postoperative hemorrhaging occurred in one patient with a MELD score of 12. Therefore, the rate of multi-complications was higher in patients with a MELD score higher than 11.5 than in those with a score lower than 11.5 (10.6% vs.0%, χ2= 3.413, P= 0.035). The rate of intraoperative hemorrhage was higher in patients with a MELD score higher than 11.5 (21.3% vs.0%, χ2=5.574, P=0.018). Additionally, the average lengths of postoperative stay were 8.0 ± 5.1 days (range from 1 day to 24 days). And the length of stay was significantly longer in patients with a higher MELD score (F=2.596, t= 1.600, P=0.035). ROC analysis identified MELD-Na scores higher than 20.9 as the best cutoff value for predicting the incidence of complications (AUC=0.74,95% confidence interval (CI)=62.3%-86.5%; sensitivity=58.3%; specificity=93.0%, P =0.001). The rate of total complications was higher in patients with a MELD score higher than 20.9 than those with a score lower than 20.9 (60.9% vs.18.5%,χ2= 10.266, P=0.000). And the other complications, included biliary tract, hepatic encephalopathy and respiratory failure were all occurred in patients with high MELD-Na scores (13.0% vs.0%, P=0.024). But there was no significant difference between the high scores and low scores according to the transfusion, multi-complications, intraoperative hemorrhaging, postoperative hemorrhaging and length of postoperative stay.ROC analysis identified Child-Pugh scores higher than 6.5 as the best cutoff value for predicting the incidence of complications according to female patients (AUC=0.74,95% CI=53.1%-89.2%; sensitivity=80.0%; specificity=73.1%, P= 0.052),8.5 as the best cutoff according to patients without jaundice (AUC=0.72,95% CI=58.3%-85.9%; sensitivity=50.0%; specificity=82.9%, P=0.014) and 8.5 as the best cutoff according to no-hepatitis B cirrhotic patients (AUC=0.72,95% CI= 58.3%-85.9%; sensitivity=50.0%; specificity=82.9%, P=0.014). Also the AUC were 0.69 according to male patients, but Child-Pugh scores had no prognostic value according to male patients, patients with jaundice and hepatitis B cirrhotic patients.ROC analysis identified MELD scores higher than 15.6 as the best cutoff value for predicting the incidence of complications according to male patients (AUC =0.77,95% CI=63.0%-91.5%; sensitivity=78.6%; specificity=70.4%, P= 0.082),11.5 as the best cutoff according to patients with jaundice (AUC=0.79,95% CI=64.2% - 93.5%; sensitivity = 78.6%; specificity=65.9%, P=0.070)，15.5 as the best cutoff according to hepatitis B cirrhotic patients (AUC=0.72, 95% CI=55.4%-89.0%; sensitivity = 64.3%; specificity=67,9%, P=0.087) and 8.5 as the best cutoff according to no-hepatitis B cirrhotic patients (AUC=0.77, 95% CI=60.8%-93,8%; sensitivity = 88.9%; specificity=62.5%，P=0.064). Also the AUC were 0.68 according to female patients, but Child-Pugh scores had no prognostic value according to female patients and patients with jaundice.ROC analysis identified MELD-Na scores higher than 20.9 as the best cutoff value for predicting the incidence of complications according to patients without jaundice (AUC = 0.75, 95% CI=62.6%-86.7%; sensitivity=58.3%; specificity=84.9%, P=0.001), 20.6 as the best cutoff according to male patients (AUC=0.75，95% CI=59.3%-90.0%; sensitivity=71.4%; specificity=74.1%, P=0.011)，18.5 as the best cutoff according to female patients (AUC=0.71, 95% CI=50.2%-90.9%; sensitivity= 50.0%; specificity=84.6%, P=0.059)，20.9 as the best cutoff according to hepatitis B cirrhotic patients (AUC=0.75，95% CI=58.3%-91.0%; sensitivity=57.1%; specificity=85.7%, P=0.001) and 11.8 as the best cutoff according to no-hepatitis B cirrhotic patients (AUC=0.75, 95% CI=56.9%-93.1%; sensitivity=88.9%; specificity=54.2%, P=0.029). But MELD-Na scores had no prognostic value according to patients with jaundice.Twelve (15.6%) patients were lost to follow-up, and the Kaplan-Meier curves showed the 5-year survival rate was 47% for all patients and the median survival time was 5.0 years (95%CI=4.1-5.9). The 5-year survival rate was 36.8% for male patients, and 59.9% for female patients. The median survival time had statistically significant difference between male and female patients (5.0 vs. 9.0,χ2=10.266, P= 0.005). But there was no statistically significant difference according different etiology, complication (yes or no) or EST (yes or no)(P=0.918, 0.068，0.086, respectively). The 5-year survival rate was 41.1% for patients with Child-Pugh A, 51.7% for patients with Child-Pugh B, and 44.4% for patients with Child-Pugh C. And the median survival time had statistically significant difference in patients with different Child-Pugh classifications (χ2=1.700, P=0.427). Furthermore, there was also no statistically significant difference for paired comparison according to Child-Pugh A, B or C. The 5-year overall survival rate in patients with higher MELD scores (>11.5) and lower scores were 43.9% and 51.6%, respectively. The median survival time was 6 years (95% CI=4.1-7.9) in patients with lower scores and 5 years (95% = CI,4.2-5.8) in patients with higher scores (χ2= 0.735, P=0.391). The gender, etiology, complication (yes or no) or EST (yes or no) did not affect the prognostic value of survival time in both Child-Pugh scores and MELD scores. The 5-year overall survival rate in patients with higher MELD-Na scores (>20.9) and lower scores were 20.1% and 60.3%, respectively. The median survival time had a statistically significant difference between patients with higher MELD-Na scores and lower scores (4.0 vs.7.0, χ2= 9.919, P=0.002). And the prognostic value of survival time was not affected by male patients, no-hepatitis B cirrhotic patients or patients without EST, but no prognostic value for hepatitis B cirrhotic patients or patients with EST.Conclusion1. ERCP with EST/EPBD can be considered an effective and safe therapeutic method in cirrhotic patients with choledocholithiasis regardless of Child-Pugh classification. Sufficient preoperative discussion and preparation for an individual may reduce the risk of complications.2. Maybe MELD scores and MELD-Na scores but not Child-Pugh scores can predict the risk of operative complications.3. Maybe MELD-Na scores have the prognostic value of survival time for patients with choledocholithiasis after ERCP, but MELD scores or Child-Pugh scores have no prognostic value.