AI Guidance for Biopsy in Suspected Cholangiocarcinoma

Efficacy of Artificial Intelligence Aid-digital Single-operator Cholangioscopy (DSOC) Guided-biopsy Sampling in Suspected Cholangiocarcinoma: A Prospective, Randomized Trial

Digital single-operator cholangioscopy (DSOC) has emerged as a medical advance with an important role in the evaluation of indeterminate biliary lesions. This technique has demonstrated higher sensitivity in the guidance for tissue acquisition when compared with standard endoscopic retrograde cholangiopancreatography (ERCP). DSOC-guided biopsy is considered technically safe and successful for tissue collection. Hand in hand with the development of more precise diagnostic techniques, comes the implementation of artificial intelligence (AI) for diagnostic assessment. For the past decade, the role of artificial intelligence (AI) has been increasing at a rapid pace. In the biliary tract, different models have been proposed for the characterization of malignant features. Nevertheless, to date, the discrepancy between the visual impression of the operator and the histological results obtained by cholangioscopy still present, affecting the accuracy the diagnosis. Based on the above, the investigators aim to assess the diagnostic accuracy of AI for the guidance of tissue acquisition with DSOC compared to DSOC without AI for suspected cholangiocarcinoma. As a secondary aim, the investigators pursue to compare quality of AI-guided biopsies samples vs. DSOC biopsies without AI.

The diagnosis and management of biliary malignancy currently represents a medical challenge. To date, DSOC has demonstrated high sensitivity in the detection of malignant biliary lesions, nevertheless there is not a universal expert consensus for the characterization of this lesions. Also, DSOC has shown to be safe and successful for specimen collection with higher sensitivity when compared with standard ERCP. Moreover, most of the AI models proposed for characterization of neoplastic features in biliary lesions have demonstrated high reliability during DSOC performance. A model was the proposed by investigators in Ecuador, focused on the identification of features of malignancy. The detection is performed by surrounding the suspected lesion in a bounding box. The detected area is displayed in the right side of the screen. Also, the box/image of the presumptive lesion can also be recorded and reviewed afterwards. After the AI model detects the "malignant area", a tissue sample is collected and taken for histopathological studies. In addition, due to a variation of the endoscopists´ intra and interobserver agreement and the discrepancy between the visual impression and histopathological findings, the investigators intend to take advantage of our AI model as a diagnostic tool for a more precise acquisition of tissue in lesions suggestive of malignancy during real-time DSOC.

This group is comprised by patients with suggestive malignant biliary lesions assessed by DSOC for biopsy. In this group, the investigators aim to use as a complement tool an AI model for the detection of features suggestive of malignancy to perform the biopsy on the detecting bounding box signal. A further follow-up of 6 months is necessary for a confirming diagnosis of neoplastic lesions.

This group is comprised by patients with suggestive malignant biliary lesions assessed by DSOC for biopsy without AI guidance. A further follow-up of 6 months is necessary for a confirming diagnosis of neoplastic lesions.

Patients with a presumptive diagnosis of biliary malignancy will undergo DSOC + Artificial intelligence model (AIWorks) guidance for detection of neoplastic lesion during real-time procedure, tissue sampling acquisition, and histopathological analysis.

Patients with lesions suggestive of malignancy will undergo DSOC without AI guidance for sampling. Based on the observer´s criteria regarding areas suggestive of malignancy, the collected tissue sample will be sent for histopathological studies.

To confirm the diagnosis based on pathology results from specimens obtained through DSOC (with or without AI-guided biopsy) or findings from further indicated procedures, including brush cytology fluoroscopy-guided biopsy, endoscopic ultrasound-guided tissue sampling, and surgical samples. Finally, the gold standard is a six-month follow-up compared against the AI model (group 1) or the DSOC endoscopist experts' classification. The data will be verified through a 2 x 2 contingency table.

Four biopsies will be performed per each case. Rate of insufficient samples by each study group will be recorded and compared.

Inclusion Criteria: Patients referred to our center with an indication of DSOC due to suspicion of CBD tumor or indeterminate CBD stenosis. Patients who authorized for DSOC-guided biopsy. Exclusion Criteria: Any clinical condition which makes DSOC inviable. Patients with more than one DSOC. Lost on a six-month follow-up after DSOC.

Saraiva MM, Ribeiro T, Ferreira JPS, Boas FV, Afonso J, Santos AL, Parente MPL, Jorge RN, Pereira P, Macedo G. Artificial intelligence for automatic diagnosis of biliary stricture malignancy status in single-operator cholangioscopy: a pilot study. Gastrointest Endosc. 2022 Feb;95(2):339-348. doi: 10.1016/j.gie.2021.08.027. Epub 2021 Sep 8.

Robles-Medranda C, Oleas R, Sanchez-Carriel M, Olmos JI, Alcivar-Vasquez J, Puga-Tejada M, Baquerizo-Burgos J, Icaza I, Pitanga-Lukashok H. Vascularity can distinguish neoplastic from non-neoplastic bile duct lesions during digital single-operator cholangioscopy. Gastrointest Endosc. 2021 Apr;93(4):935-941. doi: 10.1016/j.gie.2020.07.025. Epub 2020 Jul 22.

Robles-Medranda C, Valero M, Soria-Alcivar M, Puga-Tejada M, Oleas R, Ospina-Arboleda J, Alvarado-Escobar H, Baquerizo-Burgos J, Robles-Jara C, Pitanga-Lukashok H. Reliability and accuracy of a novel classification system using peroral cholangioscopy for the diagnosis of bile duct lesions. Endoscopy. 2018 Nov;50(11):1059-1070. doi: 10.1055/a-0607-2534. Epub 2018 Jun 28.

Ahmed Z, Mohamed K, Zeeshan S, Dong X. Artificial intelligence with multi-functional machine learning platform development for better healthcare and precision medicine. Database (Oxford). 2020 Jan 1;2020:baaa010. doi: 10.1093/database/baaa010.

Gerges C, Beyna T, Tang RSY, Bahin F, Lau JYW, van Geenen E, Neuhaus H, Nageshwar Reddy D, Ramchandani M. Digital single-operator peroral cholangioscopy-guided biopsy sampling versus ERCP-guided brushing for indeterminate biliary strictures: a prospective, randomized, multicenter trial (with video). Gastrointest Endosc. 2020 May;91(5):1105-1113. doi: 10.1016/j.gie.2019.11.025. Epub 2019 Nov 25.

Ribeiro T, Saraiva MM, Afonso J, Ferreira JPS, Boas FV, Parente MPL, Jorge RN, Pereira P, Macedo G. Automatic Identification of Papillary Projections in Indeterminate Biliary Strictures Using Digital Single-Operator Cholangioscopy. Clin Transl Gastroenterol. 2021 Oct 27;12(11):e00418. doi: 10.14309/ctg.0000000000000418.