Endoscopic ultrasonography versus cholangiography for the diagnosis of choledocholithiasis

Marcia Irene F. Canto, MD, MHS, Amitabh Chak, MD, Thomas Stellato, MD, Michael V. Sivak, Jr., MD Background: Choledocholithiasis is a major source of morbidity among patients undergoing cholecystectomy for symptomatic gallstones. There is no consensus on the best approach to diagnosing bile duct stones. We compared the safety, accuracy, diagnostic yield, and cost of EUS- and ERCP-based approaches. Methods: Sixty-four consecutive pre- and post-cholecystectomy patients referred for endoscopic retrograde cholangiopancreatography (ERCP) for suspected choledocholithiasis were prospectively evaluated in a blinded fashion. All were stratified into risk groups using predefined criteria. Endoscopic ultrasonography (EUS) and ERCP were sequentially performed by two endoscopists. Results: The success rates of EUS and ERCP were 98% and 94%, respectively. The accuracy of EUS for diagnosing choledocholithiasis was 94%. EUS provided an additional or alternative diagnosis to bile duct stones in 21% of patients. The complication rate of EUS was significantly lower than diagnostic ERCP. An EUS-based strategy costs less than diagnostic ERCP in patients with low, moderate, or intermediate risk. Conclusions: EUS is comparably accurate, but safer and less costly than ERCP for evaluating patients with suspected choledocholithiasis. It is useful in patients with an increased risk of having common bile duct stones based on clinical criteria and those with contraindications for or prior unsuccessful ERCP. EUS may enable selective performance of ERCP and improve the cost-effectiveness of diagnosing choledocholithiasis. (Gastrointest Endosc 1998;47:439-48.)

Approximately 10% to 15% of adult Americans or roughly more than 20 million people have gallstones.1 Choledocholithiasis may complicate symptomatic gallstone disease in 3% to 14% of patients undergoing cholecystectomy.2-6 In patients older than 60 years, bile duct stones may be present in 15% to 60% of cases.1 Despite its common occurrence and associated morbidity, the optimal approach to diagnosing and treating choledocholithiasis remains
Received July 3, 1997. For revision October 19, 1997. Accepted January 18, 1998. From the Departments of Medicine (Gastroenterology) and Surgery, University Hospitals of Cleveland–Case Western Reserve University, Cleveland, Ohio. Supported, in part, by the Olympus-American Society for Gastrointestinal Endoscopy Advanced Endoscopy Scholarship. Reprint requests: Marcia Irene Canto, MD, MHS, The Johns Hopkins Hospital, Division of Gastroenterology and Hepatology, Blalock 943, 600 N. Wolfe St., Baltimore, MD 21287-4461. Copyright © 1998 by the American Society for Gastrointestinal Endoscopy 0016-5107/98/$5.00 ϩ 0 37/1/88885 VOLUME 47, NO. 6, 1998

controversial. The inconsistent availability of expertise in endoscopic stone extraction and laparoscopic common bile duct exploration underscores the problematic nature of clearing the extrahepatic bile duct of stones in this era of laparoscopic cholecystectomy. Endoscopic ultrasonography (EUS) combines endoscopy with real-time, high-resolution ultrasound. Unlike transabdominal ultrasonography (US), it provides excellent sonographic visualization of the extrahepatic biliary tree without interference of bowel gas. The use of EUS to image the extrahepatic bile duct is relatively novel, but accumulating data indicate a potential utility for EUS in the perioperative evaluation of patients with suspected choledocholithiasis.7-11 EUS is superior to US and computed tomography (CT) for diagnosing bile duct stones.12 To compare the accuracy, diagnostic yield, and safety of EUS with cholangiography for evaluating patients with suspected choledocholithiasis, we performed a prospective, sequential, controlled, blinded

M Canto, A Chak, T Stellato, et al.

EUS vs ERCP for diagnosis of choledocholithiasis

Risk Stratification and Diagnosis of Choledocholithiasis All patients underwent US and/or CT before study entry. Ten patients had undergone one or more ERCPs at other hospitals before referral. At study entry, clinical and biochemical data and results of US, CT, or previous ERCP, if applicable, were used to stratify patients into four risk groups for the likelihood of the presence of choledocholithiasis based on a modification of previously published criteria2 (Fig. 1). “High risk” patients had to have a bile duct stone diagnosed by US or CT or at least three of the five clinical criteria listed. One of two experienced endosonographers independently performed EUS at 7.5 MHz with commercially available 360-degree, sector-scanning ultrasonic endoscope (Olympus EU-M20; Olympus America, Inc., Melville, N.Y.) without knowledge of the clinical history, laboratory data, or radiologic imaging results. After topical pharyngeal anesthesia and intravenous conscious sedation, endosonography was performed using the waterfilled balloon method of acoustic coupling. EUS was performed primarily in the left lateral decubitus position, but some patients were moved to the prone position when fluoroscopy was used to ascertain the position of the ultrasound transducer. The endosonographer examined the main duodenal papilla, head of the pancreas, and extrahepatic bile duct. All ultrasonic images were continuously recorded on videotape for documentation and review. The endosonographer diagnosed choledocholithiasis if there was a reproducible hyperechoic focus within the extrahepatic bile duct with associated acoustic shadowing. The number, location, and size of all stones seen were recorded. Any abnormalities in the pancreaticobiliary system were noted, and any malignant-appearing masses were staged by determining the depth of invasion and presence of lymph nodes. The quality of the EUS image was rated on a three-point ordinal scale (excellent, good, poor) based on image resolution and complete visualization of the extrahepatic bile duct. Diagnostic certainty of the presence or absence of choledocholithiasis at EUS was rated on a three-point scale (very certain, moderately certain, and uncertain/indeterminate). One of three experienced biliary endoscopists performed ERCP immediately after EUS using standard adult video duodenoscopes (JF100, JF130, TJF100; Olympus America). The endoscopist was blinded to the results of the EUS. Unlike the endosonographer, the endoscopist was aware of the clinical history and results of US, CT, or prior attempted ERCP. The endoscopist diagnosed choledocholithiasis if there were one or more filling defects noted during bile duct opacification that were not believed to be due to air bubbles. The number and size of filling defects and number of stones extracted after sphincterotomy were recorded. The presence of bile duct dilatation and other ampullary, pancreatic, and biliary abnormalities were noted as well. The quality of the cholangiogram and certainty of diagnosis of the presence or absence of bile duct stones were rated on the same ordinal scale used for EUS.
VOLUME 47, NO. 6, 1998

Figure 1. Stratification of patients into risk groups based on predefined clinical, biochemical, and radiographic criteria. Fever, Body temperature Ն 38° C; TBILI, total bilirubin; AP, serum alkaline phosphatase; ALT, serum alanine aminotransferase.

study comparing the results of EUS and endoscopic retrograde cholangiopancreatography (ERCP). We also compared the relative cost of EUS- and ERCPbased strategies for evaluating and treating choledocholithiasis in defined patient risk groups.
PATIENTS AND METHODS Patients This study was conducted between January and December 1994. Patients were considered for inclusion in the study if they had a history of gallstones and were referred for ERCP for suspected choledocholithiasis based on two or more of the following: (1) right upper quadrant or epigastric pain; (2) abnormal serum alanine aminotransferase (ALT), aspartate aminotransferase (AST), and/or serum alkaline phosphatase, and/or total bilirubin; (3) history of acute pancreatitis; (4) recent or current acute cholangitis; (5) biliary dilatation on transabdominal US or CT; and/or (6) choledocholithiasis diagnosed by US, CT, or ERCP previously performed at another institution. Only hemodynamically unstable patients were excluded. Seventy patients met these criteria. Of these, 6 declined participation in the study and a total of 64 patients were enrolled. All participating patients gave informed consent in writing, as approved by our hospital institutional review board.

EUS vs ERCP for diagnosis of choledocholithiasis

M Canto, A Chak, T Stellato, et al.

If ERCP was unsuccessful in a precholecystectomy patient, EUS results were compared with intraoperative cholangiography. If ERCP was not successful in a postcholecystectomy patient, ERCP was repeated or the patient was followed to January 1996. A final diagnosis of choledocholithiasis was made if stones were removed from the bile duct by perioperative endoscopic sphincterotomy or intraoperative stone extraction. The absence of choledocholithiasis was confirmed if no stones were seen in patients undergoing ERCP and/or intraoperative cholangiography, another diagnosis was made, or there was no relapse of symptoms and cholestasis during a minimum follow-up of 12 months. All minor and major post-procedure complications (i.e., all undesired outcomes) of both diagnostic and therapeutic ERCP were noted, including those that did not require hospitalization. Cost Identification Because patients enrolled in the study were not actually charged for the EUS, the cost analysis is based on cost estimates and not actual costs. However, using the actual outcomes of each diagnostic test, the relative direct costs of evaluating and treating patients with suspected choledocholithiasis using EUS- and ERCP-based strategies were identified and compared after stratifying by risk group. The former strategy used EUS as a diagnostic test followed by endoscopic sphincterotomy and stone extraction at ERCP only if stones were found at EUS. The ERCP-based strategy used ERCP as both a diagnostic and therapeutic modality. The cost analysis was based on the existing approach to the management of bile duct stones at our institution. In precholecystectomy patients, intraoperative cholangiography was not routinely performed at laparoscopic cholecystectomy. Expertise in therapeutic ERCP was available at our institution; hence, bile duct stones identified preoperatively were preferentially extracted at ERCP. If ERCP was unsuccessful and multiple/large/proximal stones were identified by EUS and/or cholangiography, open common bile duct exploration was performed. Bile duct stones diagnosed by intraoperative cholangiography were preferentially removed by ERCP after laparoscopic cholecystectomy. In post-cholecystectomy patients, ERCP was reattempted at least once if the initial procedure was unsuccessful. The direct costs associated with each type of diagnostic approach were calculated from the perspective of the third-party payer. Hospital charges and professional fees do not necessarily reflect the cost to the payer in Ohio. Hence, 1994 Medicare reimbursement figures for pertinent CPT codes for outpatient procedures were used (instead of charges) for the cost analysis. These were obtained from the University Hospitals of Cleveland and the faculty practice billing offices (University Physicians, Incorporated). The total hospital facility and professional reimbursement for EUS (CPT 43259 and 76975), diagnostic ERCP (CPT 43260), endoscopic sphincterotomy (CPT 43262), and stone extraction (CPT 43264) were used in the cost analysis. The radiology costs related to ERCP (use of fluoroscopy and radiology technicians) were not included
VOLUME 47, NO. 6, 1998

because all ERCPs were performed in the Digestive Disease Center’s fluoroscopy room without a technician. The cost of diagnosis, treatment, and hospitalization for procedure-related complications were also included, if applicable. In the cost analysis, the cost of an additional ERCP was added to the cost of an EUS if the latter was unsuccessful (n ϭ 1), falsely negative or positive (n ϭ 4), or had uncertain/indeterminate results (i.e., endosonographer unsure about the presence or absence of a bile duct stone, n ϭ 6). The cost of another diagnostic ERCP was included in the cost of an ERCP-based strategy if the initial ERCP was unsuccessful (n ϭ 4) or falsely positive or negative (n ϭ 2). The cost of endoscopic sphincterotomy was added to the cost of an ERCP-based strategy when it was actually performed because of diagnostic uncertainty (n ϭ 5) or a false-positive diagnosis (n ϭ 1). The total and average costs of an EUS- and ERCP-based strategy of managing choledocholithiasis were computed. Average cost was defined as the total cost of performing a diagnostic strategy divided by the total number of patients studied. The marginal cost or the difference between each type of diagnostic strategy was calculated for each risk group and compared. Statistical Analysis McNemar’s test was used to compare paired categorical data.13 Spearman’s rank correlation was used to determine the degree of agreement in EUS and ERCP stone size.14 Kappa scores were calculated to determine the concordance between EUS and ERCP results.15 Continuous paired data (such as the estimated total and marginal costs) were analyzed using nonparametric methods, including the Wilcoxon rank-sum test.16 A p value of 0.05 was considered significant. All statistical analyses were performed using SAS 6.10 (SAS PC for Windows; SAS Institute, Inc., Cary, N.C.).

RESULTS We studied 64 patients; 42 were women and 56% were awaiting cholecystectomy. The mean age was 53 years (range 22 to 92 years). All patients underwent both EUS and ERCP. Seventy seven percent of pre-cholecystectomy patients also had intraoperative cholangiography. The overall prevalence of choledocholithiasis by stone extraction at ERCP or cholecystectomy was 31%. The distribution of patients into risk groups is given in Table 1. As expected, the actual prevalence of bile duct stones was significantly different among the risk groups (p ϭ 0.001, Fisher’s exact test). Seventy percent of high-risk patients had bile duct stones compared with only 28% of moderate risk, none of the low risk, and 4% of indeterminate risk patients. US and/or CT diagnosed only 36% of high-risk patients with choledocholithiasis and none of the moderate and indeterminate risk patients with stones.

M Canto, A Chak, T Stellato, et al.

EUS vs ERCP for diagnosis of choledocholithiasis

Table 1. Characterization of study patients by risk group
High Number Mean age (range) (ys) Mean total bilirubin (range) (mg/dL) Number of patients with stones (%) % Patients with stones correctly diagnosed by US/CT Mean stone size (range) (mm)* 20 56 (30-92) 3.9 (0-14) 14 (70%) 36% 6.8 (2.1-12) Moderate 14 62 (31-91) 2.3 (0.2-5.6) 4 (28%) 0 6.6 (2.0-9.0) Indeterminate 22 48 (22-83) 1.2 (0.0-2.0) 1 (4%) 0 5.3 (3.5-7.0) Low 8 45 (24-73) 0.6 (0.2-1.4) 0 NA NA Total 64

19 (30%)

US, Transabdominal ultrasonography; CT, computed tomography; NA, not applicable (no bile duct stones were diagnosed in this group). *Adjusted mean stone size measured at endoscopic retrograde cholangiopancreatography not different among risk groups (p ϭ 0.79).

Table 2. Accuracy, sensitivity, and specificity of ultrasound, CT, EUS, and ERCP
US/CT Sensitivity Specificity Accuracy 20% 95% 71% EUS 84% 98% 94%* ERCP 95% 98% 97%*

stones in two high-risk pregnant patients and one indeterminate risk patient with a history of severe contrast allergy. The pregnant patients were safely carried to term with conservative management for cholecystitis. Accuracy and Predictive Value EUS diagnosed 24 stones in 16 of 19 patients with choledocholithiasis. Forty-seven percent of these patients had multiple stones. When positive, EUS detected all stones seen by ERCP. Both EUS and ERCP could detect stones as small as 2.0 mm. The average stone diameter was 5.3 and 6.9 mm by EUS and ERCP, respectively (range 2 to 20 mm). The overall concordance between EUS and ERCP diagnoses of choledocholithiasis was excellent, with an observed agreement of 92% and kappa of 0.79. With stone extraction at ERCP or surgery as the reference standards, the overall sensitivity and specificity of EUS was 84% and 98%, respectively. There was no significant difference between the overall accuracy of EUS and ERCP for choledocholithiasis (94% versus 97%, p ϭ 0.61, Table 2). In contrast, the accuracy of US/CT was only 71% because of very low sensitivity (20%). EUS imaging enabled accurate location of stones in the distal and mid– common bile duct (71%) (Figs. 2 and 3), proximal common bile duct or common hepatic duct (14%), or distal main hepatic ducts (14%). The overall positive and negative predictive value of EUS for choledocholithiasis was 94% and 93%, respectively. Accuracy and predictive value of EUS for choledocholithiasis were highest in patients without high risk (Table 3). The prevalence of choledocholithiasis in postcholecystectomy patients was only marginally higher than that in patients awaiting cholecystectomy (p ϭ 0.04) (Table 4). However, the accuracy of EUS for diagnosing bile duct stones was similar in pre- or post-cholecystectomy patients (94% vs. 93%, respectively) and significantly higher than US/CT in both groups (Table 4). Likewise, the accuracy of
VOLUME 47, NO. 6, 1998

Overall agreement between EUS and ERCP diagnoses ϭ 91.5%, kappa ϭ 0.79 (excellent). US, Transabdominal ultrasonography; CT, computed tomography; EUS, endoscopic ultrasonography; ERCP, endoscopic retrograde cholangiopancreatography. *No significant difference in accuracy.

Success Rate, Image Quality, and Diagnostic Certainty EUS and ERCP successfully imaged the extrahepatic bile duct in 98% and 94% of cases, respectively (p ϭ 0.71). EUS failed in only one case because of a sedative reaction resulting in an aborted procedure. ERCP was unsuccessful in four cases (despite precut sphincterotomy) because of periampullary diverticula or other technical problems. EUS provided excellent or good quality images in 92.0% of cases, there being no difference in quality from ERCP (91.7%, p ϭ 0.68). EUS was of excellent or good quality in five patients with indeterminate or inconclusive ERCPs that led to unnecessary endoscopic sphincterotomies. EUS and ERCP imaging resulted in a certain diagnosis concerning the presence or absence of choledocholithiasis 92% and 95% of patients, respectively (p ϭ 0.49). EUS led to certain diagnoses in four patients in whom the ERCP diagnoses was uncertain. On the other hand, ERCP provided good to excellent quality images and certain diagnoses in six cases where EUS was indeterminate or uncertain (four with chronic pancreatitis, one with acute pancreatitis, and one with previous sphincterotomy). EUS was successful in three patients with relative contraindications to ERCP. It correctly excluded

EUS vs ERCP for diagnosis of choledocholithiasis

M Canto, A Chak, T Stellato, et al.

Figure 2. A, Endoscopic ultrasound image of a 9 mm distal common bile duct stone (thick arrow, pointing left) with posterior acoustic shadowing and large periampullary diverticulum (thin arrow, pointing right) in a patient with no stone on CT. B, Corresponding magnified ERCP image of distal common bile duct, filling defect (thick arrow, pointing right), and periampullary diverticulum (thin arrow, pointing left). CBD, Common bile duct.

Figure 3. A, EUS image of a nondilated common bile duct with echogenic 7 mm (thin arrow) and 9 mm stones (thick arrow) with posterior acoustic shadowing not seen by transabdominal US. The image is reversed in left-right and up-down orientation due to scanning in the long position in the duodenal bulb. B, The corresponding ERCP image shows the same two stones (7 mm stone, thin arrow; 9 mm stone, thick arrow) in the same location. CHD, Common hepatic duct; CBD, common bile duct.

M Canto, A Chak, T Stellato, et al.

EUS vs ERCP for diagnosis of choledocholithiasis

Table 3. Accuracy and predictive value of EUS for choledocholithiasis by risk group
Risk High (n ϭ 20) Accuracy Sensitivity Specificity PPV NPV 90% 85% 100% 100% 75% Moderate (n ϭ 14) 93% 75% 100% 100% 91% Low (n ϭ 8) 100% NA 100% NA 100% Indeterminate (n ϭ 22) 95% 100% 95% 75% 100%

This error was attributable to small air bubbles and led to endoscopic sphincterotomy. EUS and operative cholangiography were negative for stones. Diagnostic uncertainty led to an additional five unnecessary sphincterotomies in patients without stones. ERCP also resulted in one false negative diagnosis in a moderate risk patient who was found to have a duct stone 1 week later when ERCP was repeated for recurrent symptoms and mild cholestasis. Diagnostic Yield US/CT diagnosed only 5 of the 19 patients with bile duct stones. All these patients were in the high-risk group (Table 1). Hence, US/CT detected only 36% of the high-risk patients with choledocholithiasis. Furthermore, US/CT missed all stones in the moderate and indeterminate risk groups. EUS correctly diagnosed 13 of 14 patients with choledocholithiasis who had falsely negative US/CT; the diagnostic yield of EUS over US/CT for bile duct stones was 93%. EUS correctly identified significant additional or alternative diagnoses in 21% of patients after US and/or CT. These included chronic pancreatitis (n ϭ 5), pancreatic pseudocysts (n ϭ 1), benign bile duct strictures (n ϭ 1), and ampullary (n ϭ 2) and pancreatic tumors (n ϭ 2). In two patients with jaundice and gallstones, EUS demonstrated a small bile duct stone in addition to a small mass in the ampulla or head of the pancreas with peritumor lymphadenopathy; these abnormalities were missed by CT. Except for the malignant lymph nodes, all pathologic abnormalities diagnosed by EUS were also detected by ERCP. Safety

PPV, Positive predictive value; NPV, negative predictive value; NA, not applicable (no patients diagnosed with choledocholithiasis).

Table 4. Prevalence of choledocholithiasis and accuracy of US/CT, EUS, and ERCP for choledocholithiasis by history of cholecystectomy
PrePostcholecystectomy cholecystectomy p (n ϭ 36) (n ϭ 28) Value Prevalence of stones US/CT accuracy EUS accuracy ERCP accuracy 19% 80%* 94% 97%‡ 43% 64%† 93% 96%§ 0.04 Ͼ0.05 Ͼ0.05 Ͼ0.05

US, Transabdominal ultrasonography; CT, computed tomography; EUS, endoscopic ultrasonography; ERCP, endoscopic retrograde cholangiopancreatography. *†The differences between US/CT and EUS accuracy in preand post-cholecystectomy patients were both significant (McNemar’s test, p Ͻ 0.05). ‡§The differences between ERCP and EUS accuracy in pre- and post-cholecystectomy patients were both not significant (McNemar’s test, p Ͻ 0.05).

ERCP for choledocholithiasis was also not different in either group of patients (97% vs. 96%, p Ͼ 0.05). EUS gave three false negative results. One occurred in a patient with previous endoscopic sphincterotomy; an 11 mm distal bile duct stone was overlooked because of the presence of air in the duct. The second false negative diagnosis occurred in a patient with a retropapillary duodenal diverticulum. The endosonographer missed a 4 mm stone in the distal bile duct, which was apparent on review of the videotape. The stone was also not seen at ERCP and was diagnosed only after endoscopic sphincterotomy. On review of the EUS tape, the stone was seen in the most distal common bile duct during the last 2 seconds of imaging. The last false negative diagnosis was due to incomplete imaging of a sigmoid-shaped bile duct packed with pigment stones. ERCP gave one false-positive diagnosis in a precholecystectomy patient with indeterminate risk.

All but one (moderately severe post-ERCP pancreatitis) of the procedure-related complications were minor. EUS had a much lower overall complication rate of 1.6% compared with ERCP (12.5%, p ϭ 0.001). The complication rate of diagnostic ERCP (9.4%) was also significantly higher compared with EUS (p Ͻ 0.05, McNemar’s test). The only EUS complication was from sedation-related hypoxemia requiring naloxone and flumazenil. There were eight ERCP-related complications: six of these (hypoxemia ϭ 1, bleeding due to precut sphincterotomy ϭ 1, pancreatitis ϭ 4) were related to diagnostic ERCP and the other two complications (bleeding due to biliary sphincterotomy) occurred during therapeutic ERCP (i.e., stones were diagnosed and extracted). The one case of minor bleeding during diagnostic ERCP occurred after needle-knife sphincterotomy in a moderate risk patient with a prior failed outside ERCP. Despite precut sphincterotomy, cannulation was unsuccessful. During a later (third) ERCP, bile
VOLUME 47, NO. 6, 1998

EUS vs ERCP for diagnosis of choledocholithiasis

M Canto, A Chak, T Stellato, et al.

duct stones were found and removed. All four patients with post-ERCP pancreatitis had purely diagnostic ERCPs and either a history of gallstone pancreatitis, previously failed ERCP, and/or an unsuccessful ERCP at our institution. None of these patients had choledocholithiasis, which was correctly excluded by EUS. Cost For all patients, the average total charge for evaluating choledocholithiasis using an EUS- and ERCP-based strategy was $754 and $1109, respectively. When the cost of diagnosing and treating bile duct stones was analyzed by risk group, the difference in cost was significantly lower for an EUSbased strategy compared with an ERCP-based strategy in low and moderate/indeterminate risk groups. The cost difference was not significant for high-risk patients because of the higher prevalence of choledocholithiasis and need for sphincterotomy and stone extraction (Table 5). DISCUSSION There is a growing interest in the use of EUS to diagnose bile duct stones. When prospectively compared with US and CT,12 EUS has sensitivity and specificity rates of 97% and 100%, respectively, for diagnosing choledocholithiasis. High accuracy rates (93% to 95%) and concordance with ERCP (91%) have also been reported elsewhere.10,17 Although EUS and ERCP are very accurate in patients with a high risk for bile duct stones,10 the potential clinical impact of EUS in these patients is not clear because the majority would require ERCP for endoscopic therapy. Using defined clinical criteria, we compared EUS and ERCP in consecutive patients with different levels of risk for choledocholithiasis who would be more representative of the general population of pre- and post-cholecystectomy patients. Our results confirm those of Amouyal et al.12 who found that EUS is potentially an excellent screening test for choledocholithiasis and is superior to US and CT. The poor performance of US and CT in our study is most likely related to the high proportion of small bile duct stones (89% Ͻ 10 mm), nondilated ducts (54%), and distal location of stones (71%). Indeed, US/CT failed to diagnose any of the bile duct stones (mean diameter ϭ 5 mm) in moderate and indeterminate risk patients. Our study also found that EUS can diagnose important coexisting or alternative pancreaticobiliary conditions not apparent on US or CT. Our blinded, sequential study enables us to make direct comparisons between EUS and ERCP success rates, accuracy, quality of images, and complication
VOLUME 47, NO. 6, 1998

Table 5. Relative cost of EUS- and ERCP-based strategies for diagnosing and treating choledocholithiasis
Average cost of EUSbased strategy High risk (n ϭ 20) Moderate or indeterminate risk (n ϭ 36) Low risk (n ϭ 8) $1772 $1125 Average cost of ERCPbased strategy $1718 $1610 Average difference in cost (ERCPEUS) –$174* $485†

$ 770



*In this group, an ERCP-based strategy cost slightly less but the difference did not reach statistical significance (p ϭ 0.45). †This difference was significant (p ϭ 0.003). ‡This difference was significant (p ϭ 0.008).

rates. Our EUS success rate of 98% is higher than ERCP and comparable with that reported by others (98% to 100%).8,12,17 The extremely low failure rate of EUS is one distinct advantage over ERCP, which has a failure rate of 5% to 10% even in experienced hands.18 Hence, EUS is an excellent alternative imaging modality to perioperative ERCP. In our study, four patients with unsuccessful ERCP benefited from EUS; one patient with bile duct stones had open cholecystectomy instead of the planned laparoscopic surgery. The other three patients without stones by EUS had no further attempts at ERCP. Furthermore, the lack of exposure to ionizing radiation and contrast make EUS the best diagnostic test for pregnant patients and those with contrast allergy. EUS was comparable with ERCP in accuracy, regardless of whether cholecystectomy was imminent or had already been performed. EUS and ERCP image quality and diagnostic certainty were also comparable in this study but the “quality” of the examinations may be more objectively assessed by blinded evaluation of videotapes and x-rays by another experienced endoscopist (not by the operator). Most importantly, EUS was significantly much safer than diagnostic ERCP. Potentially, EUS imaging could have prevented unnecessary endoscopic sphincterotomies (6 of 26 or 23%) and post-diagnostic ERCP pancreatitis in 15.6% of all patients. The fundamental problem in the diagnosis of choledocholithiasis is the difficulty in selecting patients for cholangiography using clinical, laboratory, and noninvasive imaging data. Even with a history of jaundice, pancreatitis, elevated liver function tests, or a dilated bile duct on US, ERCP is positive for choledocholithiasis in only 14% to 60% of precholecystectomy patients.3,5,21-26 Similarly, when perGASTROINTESTINAL ENDOSCOPY 445

M Canto, A Chak, T Stellato, et al.

EUS vs ERCP for diagnosis of choledocholithiasis

formed in post-cholecystectomy patients, ERCP demonstrates bile duct stones in only 18% to 50%.3,22,26 These low overall prevalence rates of choledocholithiasis, similar to that in our study (30%), underscore the need for alternative methods for evaluating patients. The consistently high rate of normal ERCPs despite careful selection of patients translates into unnecessary high cost and morbidity. ERCP would be the preferred diagnostic test of choice in high-risk patients with cholangitis and/or severe pancreatitis because therapeutic intervention improves outcome in patients found to have bile duct stones.27-29 In these patients, the benefit of ERCP outweighs the risk of ERCP-related complications. In patients at low risk for bile duct stones, the prevalence of choledocholithiasis is so small that cholangiography is unlikely to be of benefit. However, this leaves a large group of patients in which the suspicion for bile duct stones is neither high nor low (corresponding to our moderate and indeterminate risk group); it is this group that engenders the greatest controversy concerning optimal diagnostic and therapeutic management.18 Is EUS a suitable alternative to ERCP for suspected bile duct stones? Prat et al.10 recently showed that the performance of EUS prior to ERCP cost more than alternative strategies. This was most likely due to the sole inclusion of patients with “strongly suspected choledocholithiasis” in the study (equivalent to our high-risk group) and the resulting high prevalence of bile duct stones (66%) requiring endoscopic sphincterotomy and stone extraction. The prevalence of choledocholithiasis (74%) and need for endoscopic therapy were also high in our high-risk patients; therefore an EUS-based strategy provided no advantage in these patients. However, in the current study, an EUS-based strategy for managing patients with moderate and indeterminate risk resulted in significant cost savings primarily because of the selective use of ERCP for endoscopic therapy alone. We propose that EUS be the initial diagnostic test of choice in non-high-risk patients in institutions with the available expertise because of its high negative predictive value, safety, and lower cost. With a negative predictive value of 91% to 100% in low-, indeterminate-, and moderaterisk patients, EUS can accurately exclude choledocholithiasis, making ERCP unnecessary. If stones are demonstrated by EUS, ERCP with sphincterotomy and stone extraction can be performed immediately. This approach reduces the number of purely diagnostic ERCPs, the incidence of post-ERCP complications, and overall cost of management. Furthermore, preoperative EUS can identify patients in whom there may be technical obstacles to the re446 GASTROINTESTINAL ENDOSCOPY

moval of stones at laparoscopic cholecystectomy (e.g., numerous or large stones or those located proximal to the cystic duct).30 These patients might preferentially undergo preoperative endoscopic stone extraction, reserving common bile duct exploration for failed ERCP. EUS can also accurately diagnose post-cholecystectomy patients with bile duct stones and minimize the proportion of unnecessary diagnostic ERCPs (57% in our study). Our morbidity rate for ERCP (12.5%) may appear to be higher than that previously reported31 because of differences in definitions. First of all, Freeman’s study31 considered only complications that required more than one night of hospitalization. We counted all patients who developed any post-procedure adverse outcome. Furthermore, Freeman’s study31 only counted post-sphincterotomy bleeding that resulted in a greater than 2 gm/dL drop in hemoglobin or required blood transfusion. We included cases of persistent bleeding requiring endoscopic therapy. Finally, we report both diagnostic and therapeutic ERCP-related complications. Freeman’s study31 reported a morbidity rate of 9.8% for ERCP with biliary sphincterotomy and not the overall incidence of post-ERCP complications. We limited our cost analysis to the identification and comparison of costs of performing EUS versus ERCP as diagnostic tests for choledocholithiasis. Because we did not compare all potential alternatives for managing bile duct stones, it is possible that clinical strategies may prove to be more costeffective than an EUS-based strategy. These strategies include: (1) perioperative MRI cholangiography32 with ERCP, (2) perioperative ERCP with laparoscopic cholecystectomy, and (3) intraoperative cholangiography or ultrasonography33 with laparoscopic bile duct exploration. EUS has certain limitations in the management of patients with choledocholithiasis. Like ERCP, EUS imaging for choledocholithiasis is limited when it is impossible to pass the ultrasonic endoscope into the duodenum (i.e., pyloric stenosis, severe ulcer disease, and Billroth II anastamoses). Furthermore, EUS sensitivity is operatordependent and therefore accuracy rates could be different in the community and in academic medical centers with less experience in EUS. Our overall EUS sensitivity of 84% for blinded evaluation is closer to that reported in an earlier small blinded study.8 False-negative examinations may be due to misinterpretation rather than true failure of EUS imaging because two of three “missed” stones in the current study were evident on review of the videotapes. Stones can be missed at the hilum12 or proximal hepatic duct and bifurcaVOLUME 47, NO. 6, 1998

EUS vs ERCP for diagnosis of choledocholithiasis

M Canto, A Chak, T Stellato, et al. nign pancreatic and biliary lesions. Scand J Gastroenterol 1986;21(Suppl 123):119-22. Prat F, Amouyal G, Amouyal P, Pelletier G, Fritsch J, Choury AD, et al. Prospective controlled study of endoscopic ultrasonography and endoscopic retrograde cholangiography in patients with suspected common bile duct lithiasis. Lancet 1996;347:75-9. Yasuda K, Nakajima M, Kawai K. Technical aspects of endoscopic ultrasonography of the biliary system. Scand J Gastroenterol 1986;21(Suppl 123):143-50. Amouyal P, Amouyal G, Levy P, Tuzet S, Palazzo L, Vilgrain V, et al. Diagnosis of choledocholithiasis by endoscopic ultrasonography. Gastroenterology 1994;106:1062-7. Armitage P, Berry G. Statistical inference. Statistical methods in medical research. 2nd ed. Oxford: Blackwell Scientific Publications; 1990. p. 120-2. Armitage P, Berry B. Distribution-free methods. Statistical methods in medical research. 2nd ed. Oxford: Blackwell Scientific Publications; 1990. p. 417-9. Dawson-Saunders B, Trapp RG. Summarizing data. Basic and clinical biostatistics. Norwalk: Appleton & Lange; 1990. p. 58-9. Daniel W. Biostatistics: a foundation for analysis in the health sciences. New York: John Wiley & Sons, Inc.; 1987. Palazzo L, Girollet PP, Salmeron M, Silvain C, Roseau G, Canard JM, et al. Value of endoscopic ultrasonography in the diagnosis of common bile duct stones: comparison with surgical exploration and ERCP. Gastrointest Endosc 1995;42: 225-31. Cotton PB. Endoscopic retrograde cholangiopancreatography and laparoscopic cholecystectomy. Am J Surg 1993;165: 474-8. Phillips EH, Berci G, Carroll B, Daykhovsky L, Sackier J, Paz-Partlow M. The importance of intraoperative cholangiography during laparoscopic cholecystectomy. Am Surg 1990; 56:792-5. Sackier JM, Berci G, Phillips E, Carroll B, Shapiro S, PazPartlow M. The role of cholangiography in laparoscopic cholecystectomy. Arch Surg 1991;126:1021-6. Arregui M, Davis C, Arkush A, Nagan R. Laparoscopic cholecystectomy combined with endoscopic sphincterotomy and stone extraction or laparoscopic choledochoscopy and eletrohydraulic lithotripsy for management of cholelithiasis with choledocholithiasis. Surg Endosc 1992;6:5-10. Martin IG, Holdsworth PJ, Asker J, Baltas B, Glinatsis MT, Sue-Ling H, et al. Laparoscopic cholecystectomy as a routine procedure for gallstones: results of an “all-comers” policy. Br J Surg 1992;79:807-10. Rieger R, Sulzbacher H, Woisetschlager R, Schrenk P, Wayand W. Selective use of ERCP in patients undergoing laparoscopic cholecystectomy. World J Surg 1994;18:900-5. Salky B, Bauer J. Intravenous cholangiography, ERCP, and selective operative cholangiography in the performance of laparoscopic cholecystectomy. Surg Endosc 1994; 8:289-91. Van Stiegmann G, Pearlman NW, Goff JS, Sun JH, Norton LW. Endoscopic cholangiography and stone removal prior to cholecystectomy. A more cost-effective approach than operative duct exploration? Arch Surg 1989;124:787-90. Vitale GC, Larson GM, Wieman TJ, Cheadle WG, Miller FB. The use of ERCP in the management of common bile duct stones in patients undergoing laparoscopic cholecystectomy. Surg Endosc 1993;7:9-11. Neoptolemos JP, Carr-Locke DL, London NJ, Bailey IA, James D, Fossard DP. Controlled trial of urgent endoscopic retrograde cholangiopancreatography and endoscopic sphincterotomy versus conservative treatment for acute pancreatitis due to gallstones. Lancet 1988;2:979-83. Leung JW, Chung SC, Sung JJ, Banez VP, Li AK. Urgent endoscopic drainage for acute suppurative cholangitis. Lancet 1989;1:1307-9. Fan ST, Lai EC, Mok FP, Lo CM, Zheng SS, Wong J. Early GASTROINTESTINAL ENDOSCOPY 447

tion8,10 as a result of the limited depth of penetration of currently available echoendoscopes. Calcifications in head of the pancreas10 can cause multiple hyperechoic foci with marked shadowing that may obscure portions of the distal bile duct. Previous endoscopic sphincterotomy can give rise to air artifact in the distal bile duct and poor image quality.17 Occasionally, variations in biliary anatomy can make identification of the bile duct difficult and incomplete when imaging from the duodenal bulb position. In conclusion, EUS is much more sensitive than US or CT for diagnosing bile duct stones, regardless of duct or stone size. EUS is a highly accurate but less invasive and significantly much safer test than ERCP for evaluating patients with suspected choledocholithiasis. Furthermore, it can provide significant additional or alternative diagnoses. High success and low complication rates make it the ideal test in pregnant patients, those with contraindications to cholangiography, and those in whom ERCP has been unsuccessful. An EUS-based strategy for selecting patients with bile duct stones enables selective performance of ERCP for endoscopic therapy and lowers cost in patients with moderate, indeterminate, and low risk. Future studies should compare the clinical utility and cost-effectiveness of EUS with other diagnostic modalities for choledocholithiasis, such as intraoperative cholangiography, magnetic resonance cholangiography,32 and laparoscopic ultrasonography.33,34


11. 12. 13. 14. 15. 16. 17.

18. 19.

20. 21.

1. Gallstones and laparoscopic cholecystectomy. NIH Consensus Statement 1992;10:1-26. 2. Clair DG, Carr-Locke DL, Becker JM, Brooks DC. Routine cholangiography is not warranted during laparoscopic cholecystectomy. Arch Surg 1993;128:551-4. 3. Grace PA, Qureshi A, Burke P, et al. Selective cholangiography in laparoscopic cholecystectomy. Br J Surg 1993;80: 244-6. 4. Gerber A. A requiem for routine operative cholangiogram. Surg Obstet Gynecol 1986;163:363-4. 5. Leitman IM, Fisher ML, McKinley MJ, Rothman R, Ward RJ, Reiner DS, et al. The evaluation and management of known or suspected stones of the common bile duct in the era of minimal access surgery. Surg Gynecol Obstet 1993;176:52733. 6. Graham SM, Flowers JL, Bailey RW, Zucker KA, Imbembo AL. Utility of planned perioperative endoscopic retrograde cholangiopancreatography and sphincterotomy in the era of laparoscopic cholecystectomy [editorial]. Endoscopy 1992;24: 788-9. 7. Amouyal P, Amouyal G, Mompoint D, Gayet B, Pallazo L, Ponsot P, et al. Endosonography: promising method for diagnosis of extrahepatic cholestasis. Lancet 1989;2:1195-8. 8. Edmundowicz SA, Aliperti G, Middleton WD. Preliminary experience using endoscopic ultrasonography in the diagnosis of choledocholithiasis. Endoscopy 1992;24:774-8. 9. Dancygier H, Classen M. Endosonographic diagnosis of beVOLUME 47, NO. 6, 1998


23. 24.




28. 29.

M Canto, A Chak, T Stellato, et al. treatment of acute biliary pancreatitis by endoscopic papillotomy. N Engl J Med 1993;328:228-32. 30. Strasberg S, Callery M, Soper N. Laparoscopic surgery of the bile ducts. In: Cotton P, Hawes R, editors. Therapeutic biliary endoscopy. Vol. 6. Philadelphia: WB Saunders; 1996. p. 81-105. 31. Freeman MR, Nelson DB, Sherman S, Haber GS, Herman ME, Dorsher PJ, et al. Complications of endoscopic biliary sphincterotomy. N Engl J Med 1996;335:909-62. 32. Soto J, Barish M, Yucel E, Siegenberg D, Ferruci J, Chuttani R. Magnetic resonance cholangiography: comparison with

EUS vs ERCP for diagnosis of choledocholithiasis endoscopic retrograde cholangiopancreatography. Gastroenterology 1996;110:589-97. 33. Chardavoyne R, Kumari-Subhaya S, Auguste LJ, Phillips G, Stein TA, Wise L. Comparison of intraoperative ultrasonography and cholangiography in detection of small common bile duct stones. Ann Surg 1987;206:53-5. 34. Stiegmann GV, Soper NJ, Filipi CJ, McIntyre RC, Callery MP, Cordova JF. Laparoscopic ultrasonography as compared with static or dynamic cholangiography at laparoscopic cholecystectomy. A prospective multicenter trial. Surg Endosc 1995;9:1269-73.


VOLUME 47, NO. 6, 1998

Sign up to vote on this title
UsefulNot useful