Intraoperative Cholangiogram Interpretation for Laparoscopic Transcystic Bile Duct Exploration: Is Concurrence Possible?

Abstract

Background:

Laparoscopic transcystic bile duct exploration (LTCBDE) represents a secure and efficacious approach for managing common bile duct (CBD) stones, particularly in patients concurrently undergoing laparoscopic cholecystectomy (LC). The decision to proceed with LTCBDE hinges on real-time assessment of the intraoperative cholangiogram (IOC), which is highly operator-dependent and poorly interpreted. No established criteria exist to guide the evaluation of IOCs for LTCBDE. This study aims to ascertain the concordance among surgeons, experienced in LTCBDE, on critical aspects of IOC interpretation.

Methods:

A retrospective collection of IOC images of 40 patients who underwent LC, IOC, and LTCBDE at a tertiary university hospital between 2017 and 2018 was undertaken. Two hepato-pancreato-biliary (HPB) and 2 acute care surgery (ACS) specialists independently reviewed the images. They were tasked with assessing IOC features hypothesized to influence duct exploration and stone extraction complexity, along with overall suitability for LTCBDE based exclusively on the IOC images. Agreement percentages were calculated and Kappa inter-rater reliability statistics were assessed.

Results:

The overall agreement percentages concerning IOC features ranged between 52.5% and 82.5% with agreements surpassing 75% deemed robust. Maximum agreement was achieved in the interpretation of cystic duct morphology, whether straight or characterized by spiral valves. The concurrence for amenability of LTCBDE also demonstrated substantial accord among surgeons (ACS: 92.5%, HPB: 95%, total: 87.5%). Consistently high agreement was evident within specialty groups.

Conclusions:

Anatomical attributes displayed superior agreement levels, as opposed to variables necessitating measurements. We recommend the development of a structured approach for the interpretation of IOC to facilitate surgical education in LTCBDE.

INTRODUCTION

The presence of gallstones in the common bile duct (CBD), choledocholithiasis, is a common cause of morbidity worldwide. CBD stones are present in 10% to 20% of patients with cholelithiasis and are found in up to 10% of patients undergoing cholecystectomy for biliary colic.^1,2 Untreated choledocholithiasis can lead to recurrent abdominal pain, cholangitis, and pancreatitis, and is a risk factor for post-cholecystectomy bile leak.³ All patients presenting with biliary colic should be evaluated for the presence of CBD stones and duct clearance achieved to prevent complications.

Endoscopic retrograde cholangiopancreatography (ERCP) is a widely available and well-recognized technique for duct clearance and is usually performed pre- or post cholecystectomy; however, it is also associated with complications including pancreatitis, cholangitis, bleeding, and duodenal perforation, with an overall complication rate of around 10%.^4–6 Sphincterotomy also confers long-term risks of biliary reflux, ampullary stenosis, and may be associated with increased rates of pancreatic, hepatic, and biliary malignancy.^7–9 A 2-stage approach (laparoscopic cholecystectomy [LC] plus pre- or postoperative ERCP) also leads to an increased length of stay, increased cost of care, and exposure to a second anesthetic.^10–15 Intraoperative ERCP is an acceptable solution which removes the need for a second anesthetic, however, logistical factors limit its availability in routine practice.¹⁵

The alternative to endoscopic duct clearance is CBD exploration, either via an open or laparoscopic approach. The stones can be retrieved via ductotomy on the cystic duct (CD; transcystic) or CBD (transcholedchol). Intraoperative cholangiography (IOC) and laparoscopic common bile duct exploration (LCBDE) at the time of LC allow for CBD stones to be identified and managed during a single procedure, thus reducing the need for ERCP. Currently, there is no consensus on the optimal approach to treatment, however, several studies have concluded that a 1-stage approach is as effective, results in shorter hospital stay and is more cost-effective than the 2-stage approach.^12,16

Despite the clear advantages, the uptake of LCBDE remains low and highly variable both across Australia and worldwide.^17–19 This is likely due to a range of factors including lack of access to and familiarity with equipment, lack of training and maintenance of skills, increased operative time, and reliability of ERCP.²⁰

Laparoscopic transcystic bile duct exploration (LTCBDE) is a challenging procedure with a significant learning curve and is considered an advanced laparoscopic skill for surgical trainees.^1,21,22 IOCs must be simultaneously assessed for the presence of CBD stones, amenability to LTCBDE, and technical difficulty by the surgeon intraoperatively. The quality of the IOC and its interpretation is variable and highly operator dependent.^23,24 There are currently no published criteria to guide surgeons to accurately assess IOCs with respect to the amenability and success of LTCBDE.

Factors that may influence the decision to proceed to LTCBDE include the diameter of the CBD, the location of the confluence of CD and common hepatic duct (CHD), CD morphology, length of the CD from its confluence with the CHD, the number of stones in the CBD, and the size of the stones compared to the CD.²² This study aimed to assess whether these features can be reliably identified on IOC imaging, and explore the variability in the interpretation of these features between hepato-pancreato-biliary (HPB) and specialist acute care surgery (ACS) surgeons.

METHODS

Developing a Consensus on the Features of Intraoperative Cholangiogram

Our study used a modified Delphi process to systematically determine the criteria for successful LTCBDE. The Delphi method provided a structured approach for achieving consensus among experts, with 2 rounds involving 4 consultant surgeons (2 HPB and 2 ACS), each experienced in LTCBDE.

Delphi Process Implementation

The process began with a meeting, where each surgeon was asked to identify IOC features that were particularly relevant to LTCBDE. After an in-depth discussion, the panel agreed on 6 IOC features requiring further examination. These factors, which the surgeons believed were essential in evaluating the feasibility and complexity of duct exploration and stone extraction, were drawn from their clinical experience and expertise.

The second round focused on refining and defining these 6 IOC features, with particular consideration given to the difficulty associated with each factor. Through this iterative discussion, feedback from the initial round was reviewed, allowing for adjustments and a deeper examination of each feature’s role in influencing procedural complexity. With all 4 consultants in agreement, the Delphi process concluded with a consensus on 6 key IOC features hypothesized to impact the complexity of duct exploration and stone extraction.

The 6 key IOC features and their definitions are outlined in Table 1, http://links.lww.com/AOSO/A451. Example IOCs are provided in Supplemental Figures 1–7, http://links.lww.com/AOSO/A451. The IOC features hypothesized to influence duct exploration and stone extraction complexity were: CD length (<1, 1, or >1 cm from confluence with the CHD); CD morphology (straight or containing spiral valves); CD insertion (confluence of the cystic and CHDs to the left or right of the hepatoduodenal ligament [hereafter referred to as a left insertion or a right insertion], and length of the CBD after confluence of ducts [referred to as a high insertion or a low insertion]); CBD diameter (<10 or ≥10 mm); number of stones in the CBD (≤2 stones, >2, or stones above the level of the confluence of ducts/an alternative pathology) and stone-to-CD ratio (stone smaller or larger than CD diameter). Each surgeon’s overall impression of amenability to LTCBDE was also recorded as a binary response (Table 2).

TABLE 1.

Intraoperative Cholangiogram Features to Predict Amenability of Laparoscopic Common Bile Duct Exploration

TABLE 2.

Inter-Rater Agreement, by Variable and Surgeon Type

Variable	Specialty	Percentage Agreement	Within-Specialty Agreement*
Cystic duct length
<1 cm, 1 cm, >1 cm	ACS	77.5	κc = 0.440* (0.11, 0.76)
	HPB	70	κc = 0.462* (0.21–0.71)
	Total	52.5	κf = 0.408†, P < 0.001
Cystic duct morphology
Straight, spiral	ACS	90	κc = 0.795* (0.60–0.98)
	HPB	85	κc = 0.705* (0.48–0.70)
	Total	82.5	κf = 0.789†, P < 0.001
Cystic duct insertion
Right high, right low, left high, left low	ACS	77.5	κc = 0.344* (0.25–0.43)
	HPB	82.5	κc = 0.521* (0.41–0.62)
	Total	65	κf = 0.445†, P < 0.001
Common bile duct size
<10 mm, ≥10 mm	ACS	82.5	κc = 0.307* (−0.15 to 0.77)
	HPB	70	κc = 0.302* (−0.02 to 0.63)
	Total	62.5	κf = 0.326†, P < 0.001
Number of stones
≤2, >2, stones above confluence or other path	ACS	72.5	κc = 0.357* (0.03–0.68)
	HPB	85	κc = 0.585* (0.21–0.95)
	Total	62.5	κf = 0.458†, P < 0.001
Stone-to-cystic duct ratio
Smaller or larger	ACS	85	κc = 0.658* (0.39–0.92)
	HPB	80	κc = 0.574* (0.30–0.84)
	Total	67.5	κf = 0.538†, P < 0.001
Amenability to LCBDE
Yes or no	ACS	92.5	κc = 0.368* (0.29–0.43)
	HPB	95
	Total	87.5	κf = 0.161†, P = 0.003

ACS indicates acute care surgery consultants; HPB, hepato-pancreato-biliary consultants; LCBDE, laparoscopic common bile duct exploration.

P values are considered significant at the ≤.05 level.

^*Within-specialty agreement—Cohen’s Kappa (κ_c) used for all variables, with 95% confidence interval in parentheses.

^†All-surgeon agreement—Fleiss’ Kappa (κ_f) used for all variables.

Case Selection of Intraoperative Cholangiograms

A collection of IOCs was then developed from a pre-existing database of all patients admitted with International Disease Classification codes for choledocholithiasis (K80.50, K80.51, K80.30, K80.31, K80.40, K80.41), across 2 years from January 1, 2017, to December 31, 2018, at a tertiary university hospital in Perth, Western Australia. Patients who underwent LC with IOC and LTCBDE during the study period were eligible for inclusion. Patients without saved IOC images or who on review of the operation report did not undergo LTCBDE were excluded from analysis.

Review of Intraoperative Cholangiogram

De-identified IOC images of included patients were individually presented to the same 4 consultant surgeons: 2 HPB specialists and 2 ACS with experience in LTCBDE. Each surgeon assessed the 6 key IOC features and commented on whether it was amenable to transcystic LCBDE. The surgeons who evaluated the IOC images did not perform the IOCs during this time period and were blinded to patient demographics, presenting symptoms, operator, and outcomes. Institutional ethics approval was obtained (Reference GEKO 27098).

Statistical Analysis

Data were stored securely within the local hospital infrastructure. Statistical analysis was completed with assistance from an institutional biostatistician. Inter-rater reliability for each IOC feature was assessed using percentage agreement between all raters and between individual raters in groups. Fleiss Kappa was used to compare all 4 raters. Cohen Kappa was used for subspecialty groups (HPB and ACS) and associated confidence intervals. Findings were reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology guidelines.²⁵

RESULTS

Sixty-three patients were identified as having an LC, IOC, and LTCBDE for choledocholithiasis from clinical coding data during the study period. After review of the operation reports, 23 patients were excluded; 11 patients proceeded to ERCP rather than LTCBDE, 6 did not require LTCBDE as stones were flushed, 1 underwent choledochoscopy without IOC, 1 was unable to cannulate the CD, 1 was abandoned due to a theater emergency, and 3 had poor quality or unavailable IOC images (Fig. 1). Intraoperative cholangiograms for a total of 40 patients were assessed, and the results were analyzed.

FIGURE 1.

Case selection. CD indicates cystic duct; ERCP, endoscopic retrograde cholangiopancreatography; IOC, intraoperative cholangiogram; LCBDE, laparoscopic common bile duct exploration; OT, operative theatre.

The overall percentage agreement concerning IOC features ranged between 52.5% and 82.5% (Table 2), with agreements surpassing 75% deemed robust. Statistical significance was achieved for all kappa values ranging from 0.326 for CBD size to 0.789 for CD morphology (minimal to moderate agreement).²⁶ Maximum agreement was achieved in interpreting CD morphology, whether straight or characterized by spiral valves. This agreement was observed within both specialty groups (ACS: 90% agreement, HPB: 85% agreement) and across all surgeons (82.5% agreement, κ_f = 0.789). The most variably interpreted features were CD length estimated as either <1, 1, or >1 cm (ACS 77.5%, HPB 70%, total 52.5%; κ_f = 0.408) and CBD size (ACS 82.5%, HPB 70%, total 62.5%; κ_f = 0.326). Agreement was always higher within specialty groups than among all surgeons (Table 2). In comparing different specialties, confidence intervals overlap for each variable suggesting no difference in interpretation between the studied ACS and HPB surgeons, though the intervals are large and raise the potential for type II error.

The concurrence on amenability for LTCBDE demonstrated substantial accord among surgeons (ACS: 92.5%, HPB: 95%, total: 87.5%) but a low Fleiss Kappa (κ = 0.161, P = 0.003). Again high agreement was evident within specialty groups.

DISCUSSION

Our results show IOC features that involve reliable anatomical landmarks, such as CD morphology (straight or spiral), showed a higher degree of inter-rater agreement, while those dependent on estimated measurement, such as CBD diameter (<10 or ≥10 mm) and CD length (<1, 1, or >1 cm) showed less concordance. The IOC is assessed in real-time, and surgeons need to rely on visual estimates of measurements based on instrument diameters and other reference points intraoperatively. No difference in agreement was seen between speciality groups for any IOC interpretation feature, suggesting that acute care surgeons were equally proficient at disease identification in this study as HPB specialists.

Despite the overall variability in the agreement on the IOC features, the percentage agreement on amenability was strong both within the specialty and across speciality groups. We still need to explain how the features of the IOC listed influence this decision on feasibility. It is a subject being studied as a follow-up to this study.

IOC is an indicator of a high standard of surgery and is a fundamental skill of advanced laparoscopic surgeons.²⁷ Australia has a culture of routine IOC and therefore the rates of IOC are higher than internationally reported rates.^18,28 The skills required to perform and interpret IOC are fundamental in LTCBDE.

The success of LTCBDE depends on both patient and technical factors. Technical features were mostly based on case series data and expert opinion.^29,30 More recent literature has focused primarily on patient factors or a combination of both patient and technical factors to predict outcomes.^31–33 The major limitation of relying on both patient and technical factors is the lack of clear guidance offered to surgeons mid-procedure. This study aimed to investigate which technical features are most reliable to help guide a technician in assessing the feasibility of proceeding to LTCBDE mid-operation and help them prepare for potential challenges.

Strengths of this study include a systematic design and a unique inter-specialty comparison. However, a small sample size, wide confidence intervals, statistical ambiguity limit the strength of its conclusions. Percent agreement is the simplest measure of inter-rater agreement, with values of >75% demonstrating an acceptable level of agreement. The Cohen/Fleiss kappa methods of statistical analysis are attempting to deal with the agreements that may have occurred by chance. However, there is much controversy regarding the interpretation of these methods. As the raters in our study are all highly skilled, high levels of the same ratings across all the variables were observed; variability in the data was low and therefore kappa was low. It is unclear how kappa should be interpreted in situations where ratings made across raters are real rather than random.³⁴

Further research is necessary to determine the importance of each IOC feature and validate the methodology in a prospective cohort; this study provides information only on reliability of interpretation, and importantly not on predictability of outcomes.

We plan to expand the results of this study by developing and validating a scoring system in a prospective case series, investigating the relationship between IOC features and success of CBD clearance as well as LTCBDE difficulty. These results could be used to design a structured teaching program for training surgeons, similar to Kaldas et al,²⁴ or a difficulty score to act as a framework for case planning and teaching. A good example of such a scoring system was produced by Ban et al³⁵ in the development of a laparoscopic liver resection difficulty score.

CONCLUSION

The significance of IOC in facilitating successful LTCBDE underscores the critical nature of its interpretation. Immutable anatomical attributes, such as CD morphology (straight or spiral) and stone-to-CD ratio (larger or smaller) displayed superior agreement, as opposed to variables necessitating measurements such as bile duct dimensions or stone-to-duct ratio. ACS and HPB surgeons showed similar agreement in most aspects of interpretation, within specialty. The overall variability between agreement on IOC features and the decision to explore, suggests surgeons within specialties are using different criteria to assess difficulty. Further research is required to investigate which IOC features are most predictive of a difficult LTCBDE and help match surgeon experience with case difficulty to improve patient outcomes.

ACKNOWLEDGMENTS

The authors acknowledge institutional biostatistician Mr Glenn Boardman for statistical analysis and Mrs Lauren Meintjes for the superb gallbladder illustrations.