The interobserver agreement of optical features used to differentiate benign from neoplastic biliary lesions assessed at balloon‐assisted cholangioscopy

Abstract

Background and Aim Balloon‐assisted cholangioscopy allows mucosal assessment of the biliary tree with pediatric endoscopes. No validated optical criteria exist to differentiate benign from neoplastic biliary lesions. We aimed to identify, validate, and revalidate optical features differentiating benign from neoplastic biliary lesions. Furthermore, we aimed to determine whether cholangioscopic appearance allows endoscopists to accurately differentiate benign from neoplastic biliary lesions. Methods Baseline: from 44 de‐identified balloon‐assisted cholangioscopy videos, a blinded investigator analyzed potential optical features distinguishing benign from neoplastic biliary lesions. Validation: during the initial “teaching phase,” 20 endoscopists viewed video clips of 11 optical features identified in the baseline study. At the subsequent “test phase,” 20 further video clips were assessed by the endoscopists blinded to clinical details and questionnaires completed for the presence or absence of optical features, favored diagnosis and diagnostic confidence. Revalidation: The six identified optical features from the validation study with at least moderate agreement were revalidated the same way 12 months later assessing 20 new lesions. Results Baseline: 11 optical features were found to differentiate benign from neoplastic biliary lesions. Validation and revalidation: six optical features demonstrated at least moderate interobserver agreement (irregular margin, dark mucosa, adherent mucous, papillary projections, tubular, or branched/disorganized surface structures). Endoscopists correctly diagnosed lesions as benign in 89% and neoplastic in 83%. When highly confident, endoscopists correctly diagnosed 96% of benign and 87% neoplastic lesions. Conclusions Six features were validated and revalidated to differentiate benign from neoplastic biliary lesions. When highly confident with a diagnosis, endoscopists usually differentiate benign from neoplastic biliary lesions.

Identification of endoscopic predictors of biliary malignancy during digital cholangioscopy

“…Our study highlights neovascularization among the four determinants of IDBS diagnosis, with enlarged and twisted vessels being strongly associated with malignancy. The vascular criterion is also met when mucosal vegetations are present: while the presence of projections has been associated with malignant strictures, 27–29 our study highlights the importance of the vascularized nature of the villous projections when predicting neoplasia, as nonvascular projections most often reflect stent‐induced hyperplasia. As in colonic and esophageal mucosal lesions, 15 the dark color of the bile duct lining in the stricture was associated with malignancy; this correlation was also found in a study by Behary et al 28 .…”

“…The vascular criterion is also met when mucosal vegetations are present: while the presence of projections has been associated with malignant strictures, 27–29 our study highlights the importance of the vascularized nature of the villous projections when predicting neoplasia, as nonvascular projections most often reflect stent‐induced hyperplasia. As in colonic and esophageal mucosal lesions, 15 the dark color of the bile duct lining in the stricture was associated with malignancy; this correlation was also found in a study by Behary et al 28 . The presence of gallstones not visualized by ERCP and associated with a reduced risk of malignancy can be interpreted in two nonexclusive ways: (i) an uncalcified lithiasis can be analyzed via cross‐sectional imaging and during ERCP as a stricture, with cholangioscopy correcting this misdiagnosis; 30 and (ii) a nonneoplastic, inflammatory obstacle promotes lithogenic duodenobiliary stasis and/or reflux without creating a sufficient degree of obstruction to oppose lithogenesis 31 .…”

Consensus guidelines on the role of cholangioscopy to diagnose indeterminate biliary stricture

“…This was pathologically correlated with the subepithelial spreading of fibrotic tumor which resulted in greater difficulty in acquiring diagnostic tissue when compared with the other criteria (37). Notably, the subsequent validation studies demonstrated that tumor vessels, papillary projection, nodules, and infiltrative lesions had fair to moderate inter-observer agreement with Kappa value of 0.12-0.26, 0.43-0.54, 0.26, and 0.34, respectively (Table 5) (39,40). Nevertheless, the consensus panelists agreed these four criteria are important and should be assessed when visual impression is required for biliary malignancy diagnosis.…”

“…The rate of cholangitis was significantly lower in patients who received antibiotics compared to the control group (1% vs. 12.8%, p < 0.001) (42). Water irrigation during the cholangioscopy can increase intra-bile duct pressure resulting in bilio-venous reflux and bacteremia [39]. Therefore, the use of prophylactic antibiotics and ensuring adequate biliary drainage is achieved in all patients undergoing cholangioscopy is recommended to reduce the risk of procedure-related cholangitis.…”

When the balloon goes up …

“…In this issue of JGH, Behary et al identified and validated six optical features used to differentiate benign from neoplastic biliary lesions assessed at balloon-assisted cholangioscopy (BAC), 1 while Sano et al reported that therapeutic endoscopic retrograde cholangiopancreatography (ERCP) using balloon-assisted enteroscopy (BAE) for anastomotic stenosis of choledocho-/ pancreaticojejunostomy could achieve long-term success rates comparable to percutaneous transhepatic treatment or surgical reanastomosis. 2 ERCP in patients with surgically altered anatomy has lower success and higher complication rates compared to ERCP in patients with normal anatomy.…”