White matter diffusion estimates in obsessive-compulsive disorder across 1,653 individuals: Machine learning findings from the ENIGMA OCD Working Group
Jiook Cha,Bogyeom Kim,Gakyung Kim,Paul M. Thompson,Willem Bruin,Guido van Wingen,Fabrizio Piras,Fabrizio Piras,Dan J. Stein,Odile A. van den Heuvel,H. Blair Simpson,Rachel Marsh,Yoshinari Abe,Pino Alonso,Stephanie H. Ameis,Alan Antičević,Paul Arnold,Srinivas Balachander,Nerisa Banaj,Núria Bargalló,Marcelo C. Batistuzzo,Francesco Benedetti,Sara Bertolin Triquell,Jan C. Beucke,Irene Bollettini,Silvia Brem,Brian P. Brennan,Jan K. Buitelaar,Rosa Calvo Escalona,Yuqi Cheng,Ritu Chhatkuli,Ana Coelho,Beatriz Couto,Sara Dallaspezia,Benjamin A. Ely,Sónia Ferreira,Martine Fontaine,Jean-Paul Fouché,Rachael Grazioplene,Patricia Gruner,Kristen Hagen,Bjarne Hansen,Yoshiyuki Hirano,Marcelo Q. Hoexter,Morgan Hough,Hao Hu,Chaim Huyser,Toshikazu Ikuta,Anthony James,Fern Jaspers‐Fayer,Selina Kasprzak,Norbert Kathmann,Christian Kaufmann,Minah Kim,Kathrin Koch,Gerd Kvale,Jun Soo Kwon,Luísa Lazaro,Junhee Lee,Christine Löchner,Jin Lü,Daniela Rodriguez-Manrique,Ignacio Martínez‐Zalacaín,Yoshitada Masuda,Koji Matsumoto,José M. Menchón,Pedro Silva Moreira,Pedro Morgado,Janardhanan C. Narayanaswamy,Jin Narumoto,Ana Ortiz,Junko Ota,José C. Pariente,Chris Perriello,Maria Picó-Pérez,Christopher Pittenger,Sara Poletti,Eva Real,Y.C. Janardhan Reddy,Daan van Rooij,Yuki Sakai,Cinto Segalàs,Zonglin Shen,Eiji Shimizu,Venkataram Shivakumar,Carles Soriano‐Mas,Nuno Sousa,Mafalda Sousa,Gianfranco Spalletta,Emily Stern,S. Evelyn Stewart,Philip R. Szeszko,Chris Vriend,Susanne Walitza,Zhen Wang,Akihide Watanabe,Lidewij H. Wolters,Jian Xu,Kazuo Yamada,Je-Yeon Yun,Mojtaba Zarei,Qing Zhao
DOI: https://doi.org/10.21203/rs.3.rs-2749722/v1
2023-01-01
Abstract:Abstract White matter pathways, typically studied with diffusion tensor imaging (DTI), have been implicated in the neurobiology of obsessive-compulsive disorder (OCD). However, due to limited sample sizes and the predominance of single-site studies, the generalizability of OCD classification based on diffusion white matter estimates remains unclear. Here, we tested classification accuracy using the largest OCD DTI dataset to date, involving 1,336 adult participants (690 OCD patients and 646 healthy controls) and 317 pediatric participants (175 OCD patients and 142 healthy controls) from 18 international sites within the ENIGMA OCD Working Group. We used an automatic machine learning pipeline (with feature engineering and selection, and model optimization) and examined the cross-site generalizability of the OCD classification models using leave-one-site-out cross-validation. Our models showed low-to-moderate accuracy in classifying (1) “OCD vs. healthy controls'' (Adults, receiver operator characteristic-area under the curve = 57.19 ± 3.47 in the replication set; Children, 59.8 ± 7.39), (2) “unmedicated OCD vs. healthy controls” (Adults, 62.67 ± 3.84; Children, 48.51 ± 10.14), and (3) “medicated OCD vs. unmedicated OCD” (Adults, 76.72 ± 3.97; Children, 72.45 ± 8.87). There was significant site variability in model performance (cross-validated ROC AUC ranges 51.6–79.1 in adults; 35.9–63.2 in children). Machine learning interpretation showed that diffusivity measures of the corpus callosum, internal capsule, and posterior thalamic radiation contributed to the classification of OCD from HC. The classification performance appeared greater than the model trained on grey matter morphometry in the prior ENIGMA OCD study (our study includes subsamples from the morphometry study). Taken together, this study points to the meaningful multivariate patterns of white matter features relevant to the neurobiology of OCD, but with low-to-moderate classification accuracy. The OCD classification performance may be constrained by site variability and medication effects on the white matter integrity, indicating room for improvement for future research.