Introduction

Patients with suspected common bile duct stones (CBDS) are frequently encountered in clinical practice. CBDS causes jaundice, acute cholangitis, and acute pancreatitis and is often an indication for treatment even if patients are asymptomatic.1 Endoscopic retrograde cholangiopancreatography (ERCP) is the first choice for CBDS diagnosis but is associated with complications. In particular, post-ERCP pancreatitis (PEP) is a serious complication with a high mortality rate in severe cases. The incidence rate of PEP is 2.6–3.5%, with death occurring in 3% of patients with PEP.2-4 Because ERCP is a high-risk procedure, an alternative, accurate diagnostic method is required. For patients suspected of having CBDS, the recommended procedures to confirm the diagnosis are liver function tests and abdominal ultrasonography.5 However, one study reported that the sensitivity and specificity of ultrasonography were 73% and 91%, respectively; thus, the diagnostic sensitivity is insufficient.6 Therefore, computed tomography (CT) is often performed in addition to ultrasonography for CBDS. The sensitivity and specificity of multidetector multiphase CT for CBDS are 78% and 96%, respectively; however, it has a low diagnostic ability for stones <5 mm.7, 8 Endoscopic ultrasound (EUS) and magnetic resonance cholangiopancreatography (MRCP) are often performed for suspected CBDS when the CT results are negative. Many studies have reported on the diagnostic ability of EUS and MRCP. A meta-analysis showed that both EUS and MRCP have high diagnostic accuracy for the detection of CBDS.9 However, it is unclear whether either one is superior for CBDS following a negative CT. The aim of this randomized controlled trial was to compare the diagnostic accuracy of EUS and MRCP.

Methods Trial design

We performed this prospective randomized controlled trial to compare the diagnostic accuracy of EUS and MRCP in patients suspected of having CBDS. This trial was conducted at Yokohama Rosai Hospital in Japan and complied with the Declaration of Helsinki and Japanese ethical guidelines for clinical research. This trial was approved by the Institutional Review Board of the Yokohama Rosai Hospital on March 18, 2019, and was registered with the University Hospital Medical Information Network Center Clinical Trials Registry (UMIN000036357). Written informed consent was obtained from all the participating patients. The Consolidated Standards of Reporting Trials (CONSORT) guidelines were followed during this trial.

Participants

Patients who visited the Yokohama Rosai Hospital between April 2019 and January 2021 for suspected CBDS were enrolled in this trial. The diagnostic criteria and severity assessment of acute cholangitis were based on the Tokyo Guidelines 2018 (TG18).10

Patients who met the following criteria were included: 20–89 years of age. Met the diagnostic criteria for suspected or definite acute cholangitis of “mild” or “moderate” severity according to TG18 despite negative CT results for CBDS. Provided written informed consent.

The CT images were in the non-contrast axial views. They were interpreted by two designated doctors (board certified gastroenterologists of The Japanese Society of Gastroenterology with more than 20 years of experience in clinical practice).

The exclusion criteria were as follows: Patients who have an allergic reaction to metals. Patients who have iron hypersensitivity. Patients who have an altered or postsurgical upper gastrointestinal anatomy. Patients who have gastrointestinal hemorrhage. Pregnant or lactating women. Patients with severe acute cholangitis as per the TG18 guideline. Patients otherwise deemed unfit by the investigators based on their clinical judgment. Intervention and randomization

The patients who were clinically suspected of CBDS were screened to determine whether they fulfilled the study recruitment criteria. The principal investigator confirmed the eligibility of patients who were then enrolled and randomized. Masking was not applicable because this was an open-label trial. A random allocation sequence stratified by age and gender, with a block size of four, was generated by the principal investigator using a web-based software.

Patients were randomly allocated to either the EUS or MRCP group. Patients who were allocated to either the EUS group or the MRCP group underwent the assigned procedure accordingly. Patients with CBDS or sludge formation on either of the two examinations (EUS or MRCP) underwent ERCP, whereas those who had no CBDS or sludge formation on either examination underwent a second examination, which was either MRCP or EUS, distinct from the initial examination. The second examination was performed within 24 h of the corresponding initial examination. When the second examination detected CBDS or sludge formation, the patients underwent ERCP. ERCP was avoided when both the initial and second examinations did not detect CBDS or sludge formation.

EUS procedure

All patients were administered midazolam intravenously before the procedure, and their heart rate, blood pressure, and peripheral oxygen saturation were monitored during the procedure. The EUS procedures were performed by one of two experienced endosonographers who have performed more than 300 EUSs each. EUS was performed using a curved linear-array echoendoscope (GF-UCT260; Olympus Medical Systems Corp., Tokyo, Japan) paired with an ultrasound system (EU-ME2 premium; Olympus Medical Systems Corp.). After scope insertion, the common bile duct was first visualized from the duodenal bulb and then from the second portion of the duodenum.

MRCP procedure

Magnetic resonance cholangiopancreatography was performed with either a 1.5-Tesla magnet (EXCELART Vantage Powered by Atlas; Canon Inc., Tokyo, Japan) or a 3.0-Tesla magnet (MAGNETOM Skyra; Siemens Healthineers AG, Erlangen, Germany) with a surface phased-array coil. The examinations were performed on fasted patients in the supine position who drank 1200 mg of FerriSeltz powder (Otsuka Pharmaceutical Co., Ltd., Tokyo, Japan). T2-weighted and non-contrast-enhanced images were obtained and used for three-dimensional reconstruction.

The source images from a thin collimation multi-slice acquisition, three-dimensional reconstruction, axial, coronal, and sagittal oblique planes were interpreted by two designated doctors who were board certified gastroenterologists of The Japanese Society of Gastroenterology with more than 20 years of experience in clinical practice.

Trial endpoints

The primary endpoint was the diagnostic accuracy of each modality upon use for the initial examination. The secondary endpoints were the diagnostic ability (sensitivity, specificity, positive predictive value, and negative predictive value [NPV]) upon use for the initial examination, detection rate and characteristics of CBDSs in the second examination, and frequency of adverse events (AEs).

Endoscopic retrograde cholangiopancreatography is considered the gold standard for the diagnosis of CBDS. A “true positive” was defined as a stone or obvious visible sludge removed on ERCP, and a “true negative” was defined as a stone or obvious visible sludge that was not removed on ERCP. In cases where ERCP was not performed, patients who did not have recurrent cholangitis within 6 months were defined as “true negatives.”11, 12

Diagnostic accuracy measures the ability of a test to detect the presence or absence of CBDS. It was defined accordingly as the proportion of correctly classified subjects among all subjects.

The detection rate was defined as the proportion of patients who were CBDS positive in the second examination compared to the number of patients who underwent a second examination in each group.

The AEs were monitored by investigators and graded according to the Common Terminology Criteria for Adverse Events (CTCAE, National Cancer Institute, Bethasda, MD, USA) v5.0. Accordingly, the investigators recorded all data in the source materials (medical records, etc.) regarding the name of the diagnosed disease, the date of onset, the severity, whether the disease was serious or non-serious, the treatments used, and the outcome. If the AE persisted at the time of the final observation at the end of the trial, the investigators conducted follow-ups until there was a return to the baseline parameter and subsequent recovery or clinical stabilization.

Sample size calculation

At the planning stage of this trial, the primary endpoint was the CBDS detection rate in the second examination. Therefore, the sample size was calculated based on this endpoint. However, after the trial commenced, we determined that the endpoint had a risk of bias. As a result, we revised the primary endpoint to diagnostic accuracy, but the sample size was maintained accordingly.

Based on a pilot study in our hospital from November 2016 to November 2018, we estimated the detection rates to be 0% for MRCP after a negative EUS and 50% for EUS after a negative MRCP. The results showed that EUS was superior to MRCP in CBDS detection; however, the retrospective nature of the results was inherently associated with the risk of bias. Thus, we planned a prospective randomized controlled study.

To detect the rate using Fisher's exact test with a two-sided alpha error of 0.05 and a power of 0.8, it was determined that 15 patients per group were required. In this study, the second examination was performed within 24 h of the initial examination, and we assumed that the dropout rate would be very low. Thus, a target of 15 patients in each group was set.

Statistical analysis

The baseline clinical and demographic characteristics, diagnostic ability, CBDS detection rate in the second examination, and frequency of AEs were compared using Fisher's exact tests. For data regarding patient age, the Mann–Whitney U test was used. Statistical significance was set at P < 0.05. All statistical analyses were performed using EZR 64-bit (Jichi Medical University Saitama Medical Center, Saitama, Japan).

Results Patients

A total of 50 patients were enrolled from April 2019 through January 2021, and the follow-ups of the sample population were completed by February 2021 (Fig. 1). The patients were randomly allocated to the EUS (n = 28) and MRCP (n = 22) groups for their initial examinations. Two patients in the EUS group and three in the MRCP group were either lost to follow-up or discontinued intervention. Thus, 26 and 19 patients were analyzed in the EUS and MRCP groups, respectively.

Flow chart summarizing the trial protocol.

Magnetic resonance cholangiopancreatography was performed on 14 patients using 1.5-Tesla scanners and five patients using 3.0-Tesla scanners on the initial examination.

The baseline characteristics of the patients are shown in Table 1. There were no significant differences in age, sex, or severity of acute cholangitis. CBDS was not detected by ultrasonography in any of the patients. All patients who were offered ERCP underwent the procedure.

Table 1. Clinical and demographic characteristics of the included patients at baseline

EUS

n = 28

MRCP

n = 22

P-value Baseline characteristics Age, years, median (range) 72.5 (36–78) 76 (39–89) 0.32 Female, n (%) 7 (25) 7 (32) 0.75 Clinical parameters, n (%) Fever (BT >38°C) 2 (7.1) 6 (27) 0.12 Cholecystectomy status 2 (7.1) 2 (9.1) 1.00 Treatment history of CBDS 2 (7.1) 4 (18) 0.39 Evidence of inflammatory response, n (%) White blood cell (×1000/μL) <4 or >10 13 (46) 12 (54) 0.78 C-reactive protein levels (×10,000 μg/L) ≥1 21 (75) 21 (95) 0.64 Total bilirubin (mg/dL) ≥2 9 (32) 7 (32) 1.00 AST (U/L) >40 27 (96) 18 (82) 0.16 ALT (U/L) >40 27 (96) 19 (86) 0.31 ALP (U/L) >120 26 (93) 17 (77) 0.22 γ-GTP (U/L) >70 26 (93) 21 (95) 1.00 Severity of acute cholangitis, n (%) Mild 24 (86) 15 (68) 0.18 Moderate 4 (14) 7 (32) 0.18 ALP, alkaline phosphatase; ALT, alanine aminotransferase; AST, aspartate aminotransferase; BT, body temperature; CBDS, common bile duct stone; EUS, endoscopic ultrasound; MRCP, magnetic resonance cholangiopancreatography; γ-GTP, γ-glutamyl transferase. Trial endpoints

The diagnostic ability for CBDS detection based on each modality is shown in Table 2. For the primary endpoint, the accuracy was 92.3% for EUS and 68.4% for MRCP (P = 0.055).

Table 2. Diagnostic ability of CBDS detection EUS MRCP P-value Accuracy 92.3% (24/26) 68.4% (13/19) 0.055 Sensitivity 100% (9/9) 33.3% (2/6) 0.011 Specificity 88.2% (15/17) 84.6% (11/13) 1.00 Positive predictive value 81.8% (9/11) 50% (2/4) 0.52 Negative predictive value 100% (15/15) 73.3% (11/15) 0.10 CBDS, common bile duct stone; EUS, endoscopic ultrasound; MRCP, magnetic resonance cholangiopancreatography.

For the secondary endpoints, the diagnostic ability of EUS was better than that of MRCP; however, this difference was not significant, with the exception of the sensitivity. No AEs occurred in any of the patients.

The results of a subgroup analysis consisting of only patients who underwent MRCP with 1.5-Tesla scanners, which accounts for the majority of cases, for CT-negative CBDS detection are shown in Table 3.

Table 3. Subgroup analysis consisting of only patients who underwent MRCP with 1.5-Tesla scanners for CT-negative CBDS detection 1.5-Tesla MRCP Accuracy 64.3% (9/14) Sensitivity 20% (1/5) Specificity 88.9% (8/9) Positive predictive value 50% (1/2) Negative predictive value 66.7% (8/12) CBDS, common bile duct stone; CT, computed tomography; MRCP, magnetic resonance cholangiopancreatography.

The CBDS detection rates from the second examination are shown in Table 4 and Figure 2. In the EUS group, 13 patients underwent ERCP because CBDS or sludge formation was detected using EUS, and one patient declined the second examination. Consequently, 14 patients underwent MRCP for their second examination after a negative EUS result. In the MRCP group, four patients underwent ERCP because CBDS or sludge formation was detected using MRCP. Three were lost to follow-up, and one declined the second examination. Consequently, 14 patients underwent EUS as the second examination after a negative MRCP. The CBDS detection rate in the second examination was 0% for MRCP after a negative EUS and 35.7% for EUS after a negative MRCP (P = 0.041). EUS had a significantly higher detection rate than MRCP during the secondary examination. In addition, MRCP did not detect CBDS in any of the second examinations performed. In total, 23 patients who did not undergo ERCP experienced no recurrent cholangitis within 6 months.

Table 4. CBDS detection rate based on the second examination MRCP after a negative EUS (n = 14) EUS after a negative MRCP (n = 14) P-value CBDS detection rate following the second examination, n (%) 0 (0) 5 (35.7) 0.041 CBDS, common bile duct stone; EUS, endoscopic ultrasound; MRCP, magnetic resonance cholangiopancreatography.

Results from the initial and second examination.

We summarized five patients in whom CBDS was detected after using EUS for the second examination (Table 5). Following EUS, one patient was diagnosed with an 8-mm CBDS, and the others were diagnosed with sludge formation. All were located in the distal bile duct. During ERCP, CBDS was confirmed in all but one patient. The EUS found either sludge or small stones, which was not detected by MRCP. Figure 3 shows a patient who was EUS-positive after a negative MRCP and underwent sludge removal using ERCP.

Table 5. Characteristics of patients with CBDS detected using EUS for the second examination Patient Age Sex Size of CBDS Location of CBDS Diagnosis of ERCP 1 87 Female One 8-mm stone Distal bile duct Normal 2 89 Male Sludge formation Distal bile duct Sludge 3 42 Male Sludge formation Distal bile duct One 3-mm stone 4 70 Male Sludge formation Distal bile duct Sludge 5 74 Male Sludge formation Distal bile duct Sludge CBDS, common bile duct stone; ERCP, endoscopic retrograde cholangiopancreatography; EUS, endoscopic ultrasound.

Common bile duct stone (CBDS) findings on endoscopic ultrasound (EUS) and endoscopic retrograde cholangiopancreatography (ERCP) in a patient after negative magnetic resonance cholangiopancreatography (MRCP). (a) MRCP could not detect CBDS. (b) EUS could detect CBDS. (c) Sludge was removed by ERCP.

Discussion

To our knowledge, this is the first prospective randomized controlled trial to compare the diagnostic ability of EUS and MRCP for suspected, but CT-negative, CBDS.

Several studies have compared the diagnostic ability of EUS and MRCP in CBDS detection. However, there is insufficient evidence regarding suspected, but CT-negative, CBDS. The American Society for Gastrointestinal Endoscopy and the European Society of Gastrointestinal Endoscopy guidelines recommend EUS or MRCP for the evaluation of patients with an intermediate risk of CBDS after a negative abdominal ultrasonography; they do not recommend CT scans.5, 13 However, the diagnostic ability of ultrasonography is insufficient, and CT is commonly used for the differential diagnosis of acute abdomen and for CBDS detection, especially in the emergency room.

Japan has the highest CT use per unit population in the world, and in clinical practice, CT is often performed to diagnose clinically suspected CBDS in patients.14 However, it is a concerning issue when CBDS is clinically suspected despite negative CT scan results. Therefore, we compared the diagnostic accuracy of the two modalities in detecting suspected, but CT-negative, CBDS.

In our study, either stones or overtly visible sludge were removed upon ERCP in 15 patients. Six (40%) patients had sludge, five (33%) had cholesterol stones, and four (27%) had pigment stones. The median stone size (interquartile range [IQR]) of nine patients was 4 mm (3–5 mm), and the median bile duct diameter (IQR) was 5 mm (4–7 mm). Small stones of size ≤5 mm could not be detected by CT.

Several studies have reported the usefulness of EUS following negative CT scans.15-17 However, it was assumed that the diagnostic ability of MRCP for CBDS was low. Kondo et al.18 reported that false-negative cases for MRCP and helical CT cholangiography had a CBDS of diameter <5 mm. Thus, small stones that cannot be detected by CT may also be missed by MRCP.

Our findings showed that the diagnostic accuracy of EUS was higher than that of MRCP, although the difference was not significant. Wee et al.19 similarly reported that 15% of the patients who had findings suggestive of CBDS but were MRCP-negative were positive for CBDS on EUS.

In our study, the CBDS that were detected by EUS but not by MRCP were either small or otherwise represented sludge formation. A meta-analysis also reported that the overall diagnostic odds ratio of EUS was significantly higher than that of MRCP because of the significantly higher sensitivity of EUS than that of MRCP, especially in the detection of small stones.20 The diagnostic sensitivity for CBDS by MRCP was relatively low (33.3%), thus its role in detecting CT-negative CBDS might be limited. The diagnostic ability of MRCP was poor only when the analysis was limited to patients who initially underwent CT.

Moreover, the NPV of EUS was high (100%). Thus, patients without CBDS by EUS may not require ERCP. This highlights that EUS plays a potential role in informing whether unnecessary ERCP can be avoided. Based on this and previous studies, EUS might be the best initial intervention for patients with suspected CBDS despite negative CT findings. EUS and MRCP have their respective advantages and disadvantages. The quality of EUS depends on the skill of the endosonographer. Conversely, MRCP can produce images of similar quality at any facility. However, factors such as claustrophobia and cardiac pacemakers preclude the application of MRCP. In selecting the diagnostic examinations, the background of each case, skill of the endosonographer, and circumstances of the facilities should be considered.

Our study had some limitations. First, the required number of participants was not appropriate for assessing the primary endpoint because we modified the endpoint following the commencement of enrollment. Based on the findings from this trial, the adequate number of participants should be determined again and further randomized controlled trials with the required number of participants should be conducted. Second, this was a single-center trial; accordingly, the external validity of this trial is limited. Therefore, a multicenter randomized controlled trial is needed. Third, this was an open-label trial, with risk of bias in diagnosis when using EUS and MRCP. Finally, the MRCP protocol quality could not be unified. The original intent was to standardize MRI using 1.5-Tesla scanners; however, this was difficult due to facility, personnel, and ethical issues. Future multicenter prospective studies are required where diagnostic performance of MRCP is standardized.

Conclusions

Endoscopic ultrasound may be superior to MRCP in its diagnostic ability to detect the presence of CBDS that are undetected on CT.

Acknowledgments

The authors would like to thank all the patients, investigators, and the institutions involved in this study.

Conflict of Interest

Authors declare no conflict of interest for this article.