Muhammed Amir Essibayi, Mohamed Sobhi Jabal, Hasan Jamil, Hamza Salim, Basel Musmar, Nimer Adeeb, Mahmoud Dibas, Nicole M Cancelliere, Jose Danilo Bengzon Diestro, Oktay Algın, Sherief Ghozy, Sovann V. Lay, Adrien Guenego, Leonardo Renieri, Joseph A. Carnevale, Guillaume Saliou, Panagiotis Mastorakos, Kareem El Naamani, Arbaz Momin, Eimad Shotar, Markus Möhlenbruch, Michael Kral, Charlotte Chung, Mohamed M Salem, Iván Lylyk, Paul M. Foreman, Hamza Shaikh, Vedran Župančić, Muhammad Ubaid Hafeez, Joshua S. Catapano, Muhammad Waqas, Muhammed Said Beşler, Yasin Celal Güneş, James D. Rabinov, Julian Maingard, Clemens M. Schirmer, Mariangela Piano, Anna Luisa Kühn, Caterina Michelozzi, Robert M Starke, Ameer E Hassan, Mark Ogilvie, Anh Nguyên, Jesse Jones, Waleed Brinjikji, Marie Teresa Nawka, Marios Psychogios, Christian Ulfert, Bryan Pukenas, Jan‐Karl Burkhardt, Thien Huynh, Juan Carlos Martínez-Gutiérrez, Sunil A. Sheth, Diana Slawski, Rabih G. Tawk, Benjamin Pulli, Boris Lubicz, Pietro Panni, Ajit S. Puri, Guglielmo Pero, Eytan Raz, Christoph J. Griessenauer, Hamed Asadi, Adnan Siddiqui, Elad I. Levy, Deepak Khatri, Neil Haranhalli, Andrew F. Ducruet, Felipe C. Albuquerque, Robert W. Regenhardt, Christopher J. Stapleton, Peter Kan, Vladimir Kalousek, Pedro Lylyk, Srikanth Boddu, Jared Knopman, Stavropoula Tjoumakaris, Hugo H. Cuellar-Saenz, Pascal Jabbour, Frédéric Clarençon, Nicola Limbucci, Vitor Mendes Pereira, Aman B. Patel, David Altschul, Adam A. Dmytriw, Ali Al Balushi, Alex Brehm, R Brehm, Kévin Premat, Justin E. Vranic, Jay A. Vachhani, Vincent M. Tutino, Mohamed K. Ibrahim, Marwa A. Mohammed, Stéphanie Elens, Julian Spears, Peter SpornsErez Nossek, Monika Killer-Oberfalzer, Mohammad Ali Aziz‐Sultan, Omer Doron
The Woven EndoBridge (WEB) device treats wide-necked bifurcation aneurysms, but occlusion rates vary. This study aims to identify factors associated with immediate WEB device occlusion. Data from patients treated with WEB devices across 36 sites were analyzed. Machine learning algorithms and ordinal regression models were developed to predict immediate incomplete occlusion for ruptured and unruptured aneurysms. The study included 1565 patients, with 436 ruptured and 1129 unruptured aneurysms. Immediate complete occlusion was achieved in 38.3% of ruptured and 32.8% of unruptured aneurysms. For ruptured aneurysms, the CatBoost classifier achieved an AUROC of 0.69. Key predictors of incomplete occlusion included pretreatment mRS, aneurysm diameter, and MCA location. Ordinal regression revealed that smoking history (OR: 1.95, p < 0.001), neck diameter (Odds Ratio [OR]: 1.50, p < 0.001), and presence of a branch from the aneurysm (OR: 2.06, p = 0.016) were associated with incomplete, while bifurcation aneurysms (OR: 0.55, p = 0.017) were associated with complete immediate occlusion. For unruptured aneurysms, the CatBoost classifier achieved an AUROC of 0.68. Significant predictors of immediate incomplete occlusion included aneurysm neck width, MCA location, and presence of daughter sac. Ordinal regression revealed that smoking history (OR: 1.29, p = 0.032), neck diameter (OR: 1.24, p < 0.001), and presence of a daughter sac (OR: 1.53, p = 0.005) were associated with incomplete, while bifurcation aneurysms (OR: 0.71, p = 0.02) and posterior circulation location (OR: 0.68, p = 0.01) were associated with complete immediate occlusion. Careful evaluation of patient demographics and specific aneurysm characteristics may help improve the outcomes of intracranial aneurysms treated with WEB device.