Endoscopy 2018; 50(11): 1049-1050
DOI: 10.1055/a-0725-8113
© Georg Thieme Verlag KG Stuttgart · New York

Cholangioscopic diagnostic classification of bile duct lesions: a worthwhile task, not an easy one

Referring to Robles-Medranda C et al. p. 1059–1070
Frederic Prat
1  Gastroenterology, Cochin Hospital, Assistance Publique-Hôpitaux de Paris, Paris-Descartes University, Paris, France
Bertrand Napoleon
2  Gastroenterology, Jean Mermoz Hospital, Lyon, Ramsay Générale de Santé, France
› Author Affiliations
Further Information

Publication History

Publication Date:
06 November 2018 (online)

Bile duct lesions are revealed by signs and symptoms that result from an obstructive process. The obstruction may be intraluminal, arising from the biliary epithelium (cholangitic or neoplastic), or extraluminal, as a result of compression by a malignancy or a regional inflammatory process such as pancreatitis. The list of diseases – benign and malignant – that can produce a biliary stenosis is pretty long, although 80 % are malignant, with pancreatic carcinoma being the most common [1]. Diagnosis is often straightforward, for example when a mass is amenable to endoscopic ultrasound-guided fine-needle aspiration [2], or when the patient’s history suggests a definite etiology (e. g. postoperative); however, some lesions are much more difficult to determine by standard means such as magnetic resonance imaging and endoscopic retrograde cholangiopancreatography with brushing or fluoroscopically guided biopsies [3].

“...the presence of “irregular or spider vascularity” should be sufficient to predict the presence of neoplasia.”

When tackling a so-called indeterminate biliary stricture (IDBS), the main challenge is therefore to exclude cholangiocarcinoma. Single-operator cholangioscopy (SOC) has been a major advance in such cases because it has made direct visualization and optically guided biopsy of these lesions technically and logistically straightforward. However, SOC-guided biopsies have shown limited sensitivity because tumor cells are often deeply embedded in a thick fibrous stroma and microbiopsies are not always able to penetrate [4]. In order to overcome this limitation, identifying the cholangioscopic features that most strongly suggest malignancy is an interesting way to improve SOC diagnostic capabilities; however, no systematic analysis of SOC findings has been conducted and published to date.

In this issue of Endoscopy, Robles-Medranda et al. present a study that included a retrospective analysis of 65 patients with suspected bile duct lesions (BDLs) followed by the prospective enrollment of 106 patients with the same inclusion criteria (i. e. suspected biliary mass or stricture with suggestive symptoms and elevated tumor markers or abnormal imaging) [5]. From the analysis of images from the retrospective patient sample, the first author defined a classification that separated neoplastic (types 1 – 4) and non-neoplastic (types 1 – 3) lesions according to the presence of villous, polypoid, inflammatory, ulcerated, flat, or honeycomb patterns; the polypoid pattern was shared by neoplastic and non-neoplastic types. The authors subsequently applied their classification to the prospective group by involving expert and nonexpert endoscopists to review subsequent SOC cases with BDL and classify their lesions accordingly. Image analysis in the retrospective dataset was based on SOC procedures performed with either the optical Spyglass DVS system or the more recent digital Spyglass DS (Boston Scientific, Marlborough, Massachusetts, USA), whereas only the latter was used in the prospective dataset. Definitive diagnosis was based on either obtaining histological evidence of neoplasia from SOC-guided biopsies or from at least 6 months’ follow-up with no evidence of malignancy. The diagnostic performance of the BDL SOC classification as well as intra- and interobserver agreements were calculated. The authors found an outstanding 96 % sensitivity, 92 % specificity, 96 % negative predictive value, and 0.04 negative likelihood ratio, with remarkable agreement between observers (94 %).

More than 10 years after the initial presentation of SOC and publication of a large number of studies supporting the role of this technique in the exploration of BDLs, particularly IDBS [4] [6], it remains a serious limitation that no standard classification of these lesions has been proposed in order to help endoscopists provide structured reports with evidence-based visual impression. Robles-Medranda et al. must be commended for their effort to offer such evidence based on a prospective collection of data from a large set of patients. Interestingly, only 12.8 % of the whole cohort had undergone previous plastic biliary stent placement at the time of SOC, which adds some value to their observations by avoiding cases with well-known stent-induced inflammatory changes. However, rather than a definitive classification of BDL and IDBS, the study has many limitations that must be considered.

The study was conducted at a single center with only one author defining the classification, and mixed images obtained from different technologies (optical and digital) were used to determine the classification. Additional methodological shortcomings must also be highlighted. First, BDLs and IDBS were not clearly defined from the outset, meaning that patients with very different types of diseases, including obviously benign ones (stones, parasites, etc.) were included. Meanwhile, some of the most carcinoma-mimicking benign diseases, such as primary sclerosing cholangitis or autoimmune cholangitis, were not included. Although it is extremely interesting to present the variety of intraductal diseases that can be observed during SOC, it is in contradiction with the judgment criteria, which clearly aim to determine a diagnostic value when neoplasia is suspected. Subsequently, only patients with suspected malignancy should have been considered for the analysis of diagnostic accuracy, whereas less than 50 % in the retrospective analysis and a little more than 50 % in the prospective group belonged to this category. Second, the observers had to analyze a subset of pictures selected by the main author, whereas an unselected video of the lesion of interest would have guaranteed a more objective assessment. Although several expert and nonexpert raters were involved in classifying BDL, the absence of a really independent expert committee is disappointing when such high kappa values are reported, in contrast to previous reports showing relatively poor agreement on rather simple and intuitive definitions of cholangioscopic features [7]. Third, it appears that neoplastic and non-neoplastic subgroup categorization has been made a priori with regard to the patient’s presentation, which is confusing as some patients with neoplasia were analyzed in the non-neoplastic group and vice versa, whereas IDBS cases were separated from suspected CBD tumors even though an IDBS is per se potentially a biliary neoplasia.

Apart from these methodological criticisms, the proposed classification could be an interesting basis from which to define, more precisely, the most efficient endoscopic criteria to identify malignancy and target biopsies. Although surface features (“flat,” “ulcer,” “polyp”) were the main criteria of interest in the proposed classification, a precise definition of these terms is lacking, and no systematic and statistical analysis of their relationship to the presence or absence of neoplasia has been undertaken. It is striking to note that all types of neoplasia in this classification have one feature in common, which is never found in non-neoplastic lesions, that is the presence of irregular or “spider” vascularity. In other words, the presence of “irregular or spider vascularity” should be sufficient to predict the presence of neoplasia. This is probably the most important finding of the study and should be more precisely analyzed in future trials, as one of the main advantages of Spyglass DS is the accurate visualization of intraductal vessels. Biliary mucosal vascular patterns can be described in great detail (size and shape, network arrangement, fragility etc.) and we think that vascular features are likely to be of great value in distinguishing benign from malignant BDLs.

In conclusion, providing an acceptable and useful classification of cholangioscopic features in BDLs or IDBS is highly desirable because biopsies do not always provide a definitive diagnosis, and the need to offer patients the right treatment and avoid unnecessary surgery is obvious. The proposed classification is a valuable step in this direction. We must continue the effort undertaken by this deserving study.