Suc B, Escat J, Cherqui D, Fourtanier G, Hay J, Fingerhut A, Millat B, for the French Associations for Surgical Research. Surgery vs Endoscopy as Primary Treatment in Symptomatic Patients With Suspected Common Bile Duct StonesA Multicenter Randomized Trial. Arch Surg. 1998;133(7):702-708. doi:10.1001/archsurg.133.7.702
To compare surgical treatment (ST) with endoscopic management (EM) in patients with suspected common bile duct stones.
Two hundred twenty eligible patients originating from 18 surgery units. Patients enrolled in this multicenter randomized study had clinical symptoms that included jaundice, mild pancreatitis (Ranson score ≤2), or mild acute cholangitis; biliary colic (with increased alkaline phosphatase levels); and common bile duct stones or a common bile duct diameter of 1 cm or larger on ultrasonography.
Two hundred two patients were randomly assigned to either ST (n=105) or EM (n=97) during a 5-year period. Both groups were comparable with respect to age, sex, American Society of Anesthesiologists score, and clinical presentation.
Main Outcome Measures
The rates of early postoperative additional procedures necessary to deal with the impossibility to perform the initial procedure, complications, and retained stones after ST or EM. Subsidiary endpoints were intention-to-treat analyses of mortality and of major complications and the duration of hospital stay.
Surgical treatment was associated with a significantly (P<.001) lower rate of 1 or 2 additional procedures (8% vs 29%) due to a significantly lower rate of the impossibility to perform the initial procedure (0% vs 5%) (P<.05), major complications (4% vs 13%) (P<.05), and retained stones (6% vs 16%) (P<.04). Minor complications occurred more often in patients having ST (4%) than in those having EM (0%) (P<.01). Cholecystectomy was performed routinely in 102 patients having ST and electively in 36 patients having EM. There was 1 death in each group initially. On an intention-to-treat analysis, 3 deaths (3.1%) occurred after EM and 1 (0.9%) after ST; this difference was not statistically significant (P=.56). Major complications occurred in 4% of patients having ST compared with 11% of patients having EM (P<.002). The median duration of hospital stay was 16 days in patients having ST and 12 days in those having EM; this difference was not statistically significant (P=.09).
Whether an additional cholecystectomy is performed routinely or electively, the high risk of additional procedures after EM precludes its use as the optimal therapy in patients with symptomatic common bile duct stones, except in those with severe cholangitis.
WHEN choledocholithiasis is associated with acute pancreatitis, a recent trial has shown that endoscopic sphincterotomy (ES) was not superior to conventional treatment, as regards overall mortality, mortality due to pancreatitis, and complications.1 In patients with severe pancreatitis, early ES (either <242 or <72 hours3) seemed to be better than late ES (>72 hours). In patients with severe cholangitis, endoscopic drainage is superior to surgery.4 In other types of symptomatic choledocholithiasis, ie, with jaundice, mild pancreatitis, mild cholangitis, or biliary colic with biochemical signs of cholestasis, 2 therapeutic options are presently available, namely, surgical treatment (ST) and endoscopic management (EM). The latter includes endoscopic retrograde cholangiopancreatography (ERCP) associated with ES and the extraction of stones whenever found. At least 6 controlled trials5- 10 have shown that immediate mortality was not significantly different, ranging from 0% to 4% in patients having ST and from 0% to 6% in patients having EM. Endoscopic management would be a valuable therapeutic alternative if the gallbladder could be left in situ.7- 9 When the gallbladder is left in situ, however, 20%7 to 40%8,9 of patients will require a second operation for biliary complications in the months or years after EM. Two operative strategies have been proposed to avoid these risks, early routine cholecystectomy either during the same hospital admission5- 7,10 or only when necessary.8,9 To choose the better of the 2 strategies, the rate and the risks of early additional procedures must be known. "Early additional procedures" are defined as those necessary to do because of the impossibility of performing the initial procedure, complications, and retained stones (including either repetition of the initial procedure or the necessity of another procedure). Additional procedures are associated with a second anesthesia and a second intervention, both representing extra costs and risks. For these reasons, it appeared necessary to compare the early results of ST and EM on an intention-to-treat basis in a multicenter trial controlled by randomization.
From September 1989 to September 1994, 220 consecutive adult (>18 years old) patients from 18 French general surgery centers (9 university hospitals, 8 teaching hospitals, and 1 private hospital) were enrolled. Each participating center included 2 to 44 patients (median, 4 patients per center) but the starting and ending dates, as well as the duration of patient inclusion, were not the same for all centers. The duration of participation ranged from 6 months to 5 years (median, 2 years). A median of 2.5 patients was included by center and per year (range, 1-9 patients). There were no restrictions as to the upper age limit, surgical risks, or previous cholecystectomy.
Patients enrolled in the study presented with 1 or more of the following: clinical symptoms that included jaundice, mild acute pancreatitis (Ranson score ≤211), or mild acute cholangitis4; biliary colic associated with increased alkaline phosphatase levels (with or without increased bilirubin level); and common bile duct (CBD) stones or a CBD diameter of 1 cm or larger on ultrasonography.
Patients who had cholecystitis (gallbladder wall >3 mm on ultrasonography) or who did not have cholelithiasis visible on ultrasonography with a CBD of less than 1 cm and patients who might not be able to undergo EM because of a previous total gastrectomy, a Billroth II distal gastrectomy, or choledochoenterostomy were not included. Participating centers were free to exclude patients if 1 of the 2 treatments was deemed preferable, as in patients with severe acute pancreatitis (Ranson score >2)2,3 or severe acute cholangitis.4
Patient characteristics, including prognostic factors according to age and American Society of Anesthesiologists (ASA) score, are shown in Table 1.
Surgical treatment included cholecystectomy (if not performed previously) and CBD exploration by intraoperative cholangiogram, choledochotomy, or both. Choledochoscopy was not mandatory. After CBD exploration was completed, the choice between primary CBD closure, T-tube drainage, or choledochoenterostomy was left to the surgeon. Retained stones were diagnosed either by postoperative T-tube cholangiograms or by ERCP performed if patients had clinical or biological signs of possible CBD stones.
In patients having EM, ERCP was performed after neuroleptanalgesia. Endoscopic sphincterotomy was performed, followed by stone basket extraction, after the confirmation of CBD stones. Endoscopy was performed by gastroenterologists and not by surgeons. Early cholecystectomy was mandatory only if acute cholecystitis or cholangitis developed. Cholecystectomy was not performed routinely during the same hospital stay after successful EM but according to each surgeon's preference.
Only patients with acute cholangitis had prolonged antibiotic therapy (>6 days), whereas the others had antibiotic prophylaxis (<24 hours).
The main outcome measure was the number of patients requiring an additional procedure as defined above. Subsidiary end points were intention-to-treat analyses of mortality and major complications, and the duration of hospital stay.
Major complications included locoregional complications amenable to additional procedures. Examples were abdominal complications requiring a second operation or EM in the patients having ST; hemorrhage, perforation, acute cholecystitis or cholangitis requiring revisional endoscopy, or ST in the patients having EM; acute pancreatitis or gastrointestinal hemorrhage in both groups; and general complications with potential fatal consequences. Examples of general complications were respiratory failure, cerebral (stroke, hemiparesis, or hemiplegia) compromise, or cardiovascular (myocardial infarction, pulmonary embolism, heart failure, or severe rhythm disorders) compromise.
Minor complications were those that healed spontaneously or after medical treatment such as wound complications in patients having ST or mild respiratory tract infections or deep venous thrombosis in both groups.
Retained stones were those resulting from incomplete or unsuccessful clearance and were always treated by either the same (essentially for EM) or another extraction technique 3 to 7 days after the last procedure was performed. The postoperative period was defined as 30 days after the initial operation.
Based on results in the literature and using the explicative, 1-tailed analysis, 95 patients per group, ie, 190 patients in all,12 were required to lower the additional procedure rate from 25%8 to 8%9 with a power of 90% (β risk, .10) at α level .05.
Treatment was randomly assigned by a telephone call to the coordinating center. Treatment assignment was according to random number tables and balanced every 4 patients in each center.
Data were collected prospectively in each participating center by filling in flow sheets, whereas analysis was centralized in the coordinating center. Comparisons were made with the χ2 and Fisher exact tests for discrete variables, analysis of variance for continuous variables, the Student t test for means, and the Mann-Whitney U test for nonparametric variables.
Of 220 eligible patients, 18 (8.2%) were withdrawn secondarily because of an incorrect diagnosis (malignant biliary obstruction in 9 patients) or protocol violation (9 patients). Of the 202 remaining patients, 105 had been randomly assigned to ST and 97 to EM.
As shown in Table 1, patients in both groups were similar with respect to age, sex, ASA score, and clinical presentation.
All 105 patients in the group having ST were operated on and had either intraoperative cholangiograms alone (n=21), choledochotomy alone (n=25), or both (n=59). Fifty-six underwent choledochoscopy. Eighty-one patients (77.1%) had CBD stones, whereas in 24 patients (22.9%), no stones were found at cholangiography or choledochotomy. Cholecystectomy was performed in 102 patients who still had their gallbladder, whereas 3 patients had undergone a previous cholecystectomy. Surgical procedures included cholecystectomy alone in 13 patients (12.4%), cholecystectomy plus T-tube drainage in 65 patients (61.9%), primary CBD closure without external drainage in 11 patients (10.5%) or choledochoenterostomy in 15 patients (14.3%) (in 2 of these, the operation was aborted without stone extraction because of frail patient condition), and surgical sphincteroplasty in 1 patient (1.0%).
Endoscopic retrograde cholangiography was successfully performed in 92 (95%) of 97 patients. It was impossible to catheterize the papilla in 5 (6% [5/97]) patients, a significantly higher failure rate than in patients having ST (0%; P<.05) (Table 2). Of these 92 patients, 80 (87%) had CBD stones and underwent an attempt at extraction, whereas 12 (13%) did not and underwent no further investigations. Retained stones were detected after ES and extraction in 16 (20%) of the 80 patients with ERCP-confirmed CBD. On an intention-to-treat basis, however, the retained-stone rate in patients having EM was 16%. This was significantly higher than in those having ST (6%; P <.02) (Table 2).
Additional procedures were required in 8 (7.6%) of 105 patients having ST compared with 28 (29%) of 97 patients having EM (P<.001). In patients having ST, this included 2 subsequent operations (2%) for hemoperitoneum and persistent jaundice (no cause found at a second operation) in 1 patient each, and 6 postoperative ES (6%) to treat retained stones. In 97 patients having EM (Table 3), 6 (6%) required revisional endoscopy for retained stones, and 20 (21%) required surgical intervention, including cholecystectomy with choledochotomy as indicated. Among these were 5 patients for whom EM was impossible (all of whom had CBD stones at surgery), 5 patients with major complications requiring a second operation (4 cases of acute cholecystitis and 1 case of cholangitis), and 10 patients with retained stones. Two further major complications that developed in patients having EM (acute pancreatitis and retroperitoneal perforation) healed without subsequent operations.
The complication rate was similar (11%) in the 2 groups. Major complications occurred less frequently after the first procedure in patients having ST (Table 1) (4% vs 11%; P <.05) and in intention-to-treat analysis (4% vs 13%; P<.02) (Table 3). Minor complications, however, occurred more often in patients having ST (7% vs 0%; P <.01).
As shown in Table 2, there was 1 death in each group after the initial treatment. An 89-year-old woman in the group having ST with an ASA score of 3 and with mild cholangitis died of pneumonia after choledochoenterostomy. Another 83-year-old woman with an ASA score of 3 and with mild acute pancreatitis died of stroke after successful EM.
After secondary treatment (Table 2), an 88-year-old woman with an ASA score of 2 and with isolated jaundice died of myocardial infarction after a second, unsuccessful clearing of residual stones.
Routine secondary cholecystectomy was performed in 36 patients (37%) (21 by an open procedure and 15 by laparoscopy) between 2 and 20 days (median, 5 days) after EM. One of these patients, an 81-year-old man with an ASA score of 2 and with mild cholangitis, died of abdominal hemorrhage after a routine cholecystectomy performed 5 days after successful EM. A second patient had major complications (hematoma and burst abdomen) and recovered uneventfully after a subsequent operation.
Analyzed with intention to treat, deaths occurred less often in patients having ST (1%) than in those having EM (3%), but this difference was not statistically significant (P=.56). All deceased patients had ASA scores of 2 or 3 and were older than 80 years. The median hospital stay was 4 days longer in patients having ST than in those having EM (P=.09).
Comparisons between participating centers did not show any significant differences in the selection of patients or in outcome.
As an initial procedure, ST was significantly associated with fewer additional procedures (P<.001), fewer retained stones (P<.002), fewer major complications (P<.05), but more minor complications (P<.01); when analyzed with intention to treat, ST was associated with less mortality (P=.56), and a longer hospital stay (P=.09) (Table 2 and Table 3).
The lack of uniformity of surgical and endoscopic techniques and perioperative or periprocedural care is a limitation of this study. The population in this surgical center study probably does not reflect the general population of patients with symptomatic CBD stones and notably that of patients seen and treated by gastroenterologists only. That these patients were referred to surgical centers most likely indicates that they were more difficult to treat. Nevertheless, because this was a multicenter study involving both gastrointestinal endoscopic specialists and surgeons, the extrapolation of results to all centers involved with the treatment of this same type of disease is possible.
In our series, EM was impossible to perform in 6% of patients. In the 6 controlled trials to date, this rate ranged from 0% for the smaller series6,8 to 3%,10 4%,5,7 and 10%.9 This rate was higher in series of patients with acute pancreatitis only: 11%4 and 12%.1,3
The rate of retained stones should be calculated according to the number of patients with CBD stones that require extraction. In some studies,5,7,8 patients were enrolled only after the diagnosis of CBD stones through ERCP; understandably, all patients had calculi. In other studies,6,9,10 patients were enrolled when stones were simply suspected, and then the rate of stones after successful EM ranged from 50% to 80%, as in our series (Table 1). Furthermore, the successful extraction rate should be calculated according to the first attempt at EM because ulterior attempts at extraction are in fact additional procedures. The rate of retained stones in our series after initial EM was 20%. This is the median value found in the literature, in which the rate ranged from 4%,9 13%,5 17%,7 23%,8 to 29%.6,10 In some series, this rate was calculated after spontaneous clearance of the CBD stones following ES.7,8 Moreover, clearance of the CBD stones required additional diagnostic procedures such as ERCP that were not accounted for in these studies. Two reports5,9 did not mention second attempts at EM, but, instead, ST was decided, whereas in other trials,6- 8,10 a second attempt at EM was made before routine cholecystectomy. The final median rate of retained stones after a second endoscopy was 16% (the same as in our series [Table 1]), ranging from 6% to 25%.6- 8,10 The spontaneous evacuation of stones may occur between the time of ES and surgical procedure5- 7 and even if ES was not performed.13,14 In patients with mild acute pancreatitis11 or cholangitis1, however, the early clearance of CBD stones is necessary to avoid infection (cholecystitis and cholangitis) that may occur after endoscopic investigations.7,15 At present, the clearance rate of stones can be increased by endoscopic lithotripsy,15 whereas the complete clearance of multiple or large calculi remains unpredictable.7
The rate of retained stones after ST was 6% in our series. In other series, this ranged from 2% to 14%.5- 10 In these trials, choledochoscopy was either not mentioned or not used, whereas in our series, 56 choledochoscopies were performed in 81 patients with CBD stones, a possible explanation for our low retained stone rate, as also shown elsewhere.16 If choledochoscopy had been performed routinely in our series, however, the incidence of retained stones would probably have been even lower.
Acute cholecystitis (3% in our series) leading to emergency surgical intervention was triggered by endoscopic investigations,7 whereas cholangitis was most always due to retained stones.7,15 In controlled trials without early routine cholecystectomy,8,9 these complications have been reported to range from 0% to 10% (median, 5%). In our series, routine antibiotic therapy was unable to prevent or eradicate any of these complications.
Additional procedures have not always been mentioned or defined in the literature. In series including planned cholecystectomy, the rate of necessary additional procedures can, however, be deduced to be at least 18%6,7 to 24%10 to 35%.5 In contrast, when early cholecystectomy was not planned, additional procedures were performed in 10%9 to 23%8 of patients. Our 29% rate is a bit higher, confirming the value of taking the additional procedures into account. In patients with EM, 37% had routine cholecystectomy, probably because many surgeons thought that cholecystectomy was not necessary after ES. The only controlled study suggesting the contrary was published in 1995,8 1 year after the end of our study. It is possible that an intact gallbladder could contribute to a high incidence of retained CBD stones and additional procedures.
Only 2 of the studies published to date9,10 analyzed results based on the intention to treat. The most commonly reported end points have been complication and mortality rates9 and length of hospital stay10 with regard to the initial treatment scheme. Each additional procedure performed, however, whether surgical or endoscopic, is associated with additional risks. Obviously, cumulative morbidity and mortality based on 2 or 3 procedures might be higher than those for 1 procedure.7 This is well documented in our study because, of 2 patients initially included in the group having EM, 1 died after a second attempt at EM and the other, already mentioned, died after early routine secondary cholecystectomy.
Morbidity is not defined or expressed in the same manner throughout the literature. In our series, the rate of major complications was significantly higher in patients having EM, whereas the rate of minor complications was significantly higher in those having ST. This was found in 3 other trials in the literature7- 9 in which the major complication rate was less in patients having ST: 7%, 8%, and 9% than in patients having EM: 8%, 10%, and 25%, respectively. On the other hand, minor complications occurred less often in patients having EM, 6%9 and 10%,8 than in those having ST (15% for both). In 2 trials,6,10 overall morbidity was higher in patients having EM: 12%10 and 21%,6 than in patients having ST: 10% and 18%, respectively. In one study,5 the rates of major (12%) and minor (8%) complications were the same in both groups. In conclusion, we agree with Neoptolemos et al7 that complications were significantly related to the failure of initial procedures.
Compared with that of EM, the rate of immediate death after ST in our series was lower, but not statistically significantly so. In the literature,5,6,8 mortality was nil in both groups of patients when the series was small, ie, 34, 52, and 80 patients. Of note, however, none of these series had determined the number of patients required with a preset β error. In larger series, however, including 98,9 115,7 and 207 patients,10 mortality, contrary to popular belief, was always higher in patients undergoing EM than in those undergoing ST (6% vs 4% [P=.98], 4% vs 1% [P=.91], and 2% vs 1% [P=.95], respectively), even though these differences were not statistically significant. If the 2% difference in mortality found in our study were truly the case, more than 2000 patients (in a 2-tailed analysis) would have been required to prove this with α and β errors of .05. This possible difference in favor of ST has wide practical implications in view of the large number of endoscopies performed for CBD stones in everyday practice.
As a 2-step procedure, routine cholecystectomy carries an attending mortality rate that has to be added to that of EM alone (1 death in 36 secondary cholecystectomies in our series).
It is commonly thought that EM is less invasive in elderly and high-risk patients. As in the series of Targarona et al,9 EM was associated with more severe complications (P=.97) and higher mortality (P=.98) than ST in elderly and high-risk patients, as in our series (Table 1), but these differences were not statistically significant. In other series, aged or high-risk patients were not studied separately, whereas the European Association of Endoscopic Surgeons study10 did not include patients with ASA scores of 3 or 4.
In controlled trials,5,7- 9 the hospital stay was longer in patients having ST than in those having EM with or without routine cholecystectomy, as in our series (Table 3). In 2,6,10 and especially 1 trial,10 where all surgical procedures were performed by laparoscopy, however, the cumulative duration of hospital stay in patients having EM plus routine cholecystectomy was 2 to 3 days longer than for those having ST alone (P=.05). The median hospital stays for EM in the literature were 4,5 9,7 9½ ,10 and 13 days.8 Our median hospital stay of 12 days (Table 3) for EM may seem long. Possible explanations might be found in the surgical referral pattern used in this study, the age of patients and/or the inclusion of patients with ASA scores of 3 to 4 (50% of patients), and the rates of early additional procedures (29%) and early routine cholecystectomy (36%) (Table 3). Conversely, the difference in the duration of hospital stay in favor of EM in our series stems from the fact that 34% of patients did not have early associated cholecystectomy, whether performed routinely or only when necessary (Table 3).
On the one hand, as shown elsewhere,7- 9 between 20% and 40% of patients with gallbladder left in place will require a second operation during the months or years that follow EM, and as 25% of patients undergo an early additional procedure, this means that more than 50% of patients will have a second operation. On the other hand, the rate of additional procedures can exceed 100% if a second endoscopy becomes necessary in patients initially treated with EM followed by early routine cholecystectomy.
The rate of second anesthesia for additional procedures and, consequently, the additional risks and costs are such that EM alone is insufficient and not warranted in patients with symptomatic choledocholithiasis who have not had cholecystectomies. The only indication for initial EM would be the case of a patient with a previous cholecystectomy because in that case, the risks related to leaving the gallbladder in place are eliminated. Surgical treatment is more advantageous than EM because the gallbladder can be removed (thus eliminating the risk of subsequent acute cholecystitis) and the CBD visualized directly by choledochoscopy. Routine combined endoscopic and surgical treatment cannot be the choice for CBD and gallbladder stones nowadays because of the increased risks and costs associated with more than 1 anesthesia and additional procedures.
Studies targeting quality of life and costs with intention-to-treat analysis may still be necessary before definitively proposing ST as the standard of care in these patients. Traditional surgical therapy can be replaced advantageously by laparoscopic cholecystectomy and laparoscopic treatment of choledocholithiasis whenever necessary, as shown in the randomized study of the European Association of Endoscopic Surgeons.10 This requires adequate experience and skill, however.17
We thank the following surgeons for participating in this study: Christian Dilin, MD (Thonon-les-Bains); Pierre-Louis Fagniez, MD; Nelly Rotmann, MD (Créteil); Jean-Charles Etienne, MD (Poissy); Philippe Gabelle, MD (Grenoble); Alain Gainant, MD (Limoges); Yves Flamant, MD; Guy Zeitoun, MD (Colombes); Daniel Jaeck, MD (Strasbourg); Gérard Kolhmann, MD (Corbeil-Essonnes); Jean-Pierre Lenriot, MD; Jean-Christophe Paquet, MD (Longjumeau); Christian Letoublon, MD (Grenoble); Patrice Le Treut, MD (Marseille); Jacques Marescaux, MD (Strasbourg); Philippe Marre, MD; Thierry Montariol, MD (Saint-Germainen-Laye); (Montpellier); Simon Msika, MD (Meulan); Jean-Louis Pailler, MD (Neuilly-sur-Seine); Jacques Testard, MD (Rouen); and Michel Veyrières, MD (Pontoise).
Corresponding author: Abe Fingerhut, MD, Centre Hospitalier Intercommunal, Chemin du Champ-Gaillard, 78303 Poissy, France.
Reprints: Bertrand Suc, MD, Service de Chirurgie Digestive, Hôpital Rangueil, 31054 Toulouse Cedex, France.