Blood bile acid profiles in chronic inflammatory demyelinating polyneuropathy.

作者信息

Chang Chun-Wei, Cheng Mei-Ling, Chen Chiung-Mei, Liu Tsai-Wei, Ro Long-Sun, Lo Yen-Shi, Lyu Rong-Kuo, Kuo Hung-Chou, Liao Ming-Feng, Chang Hong-Shiu, Huang Ching-Chang, Wu Yih-Ru, Chu Chun-Che, Chang Kuo-Hsuan

机构信息

Section of Neuromuscular Diseases, Department of Neurology, Chang Gung Memorial Hospital-Linkou Medical Center, Taoyuan City, Taiwan; Center of Neuroimmunological and Rare Diseases, Chang Gung Memorial Hospital-Linkou Medical Center, Taoyuan City, Taiwan.

Metabolomics Core Laboratory, Healthy Aging Research Center, Chang Gung University, Taoyuan City, Taiwan; Department of Biomedical Sciences, Chang Gung University, Taoyuan City, Taiwan.

出版信息

Clin Chim Acta. 2026 Jan 1;578:120479. doi: 10.1016/j.cca.2025.120479. Epub 2025 Jul 11.

DOI:10.1016/j.cca.2025.120479

PMID:40652999

Abstract

BACKGROUND

Chronic inflammatory demyelinating polyneuropathy (CIDP) is an acquired immune-mediated neuropathy with demyelinating features like Guillain-Barréé syndrome (GBS). Despite established diagnostic criteria, the lack of specific blood biomarkers highlights the need for novel markers to improve early diagnosis and disease monitoring. Bile acids (BA), cholesterol-derived molecules with immune modulatory properties, have been explored as biomarkers in immune-mediated diseases. This study investigates BA profiles in CIDP and evaluates their potential for diagnosing CIDP.

METHODS

Patients with treatment-naïve immune-mediated polyneuropathies (CIDP and GBS) and age-matched healthy controls (HCs) were recruited from a tertiary referral hospital. Their plasma BA profiles were analyzed using liquid chromatography-mass spectrometry. A supervised machine learning model was employed to assess the BA profiles, and a simplified tree-based algorithm was developed based on the feature importance to differentiate CIDP, GBS, and HC.

RESULTS

This study included 36 CIDP patients, 70 GBS patients, and 41 HCs. Compared with HCs, CIDP patients showed elevated levels of glycochenodeoxycholic acid (GCDCA, 1,306.64 vs. 614.16 nM, P < 0.001) and glycohyocholic acid (GHCA, 16.95 vs. 6.04 nM, P < 0.001). CIDP patients also exhibited higher levels of GCDCA (1,306.64 vs. 734.44 nM, P < 0.001) and cholenic acid (8.25 vs. 4.41 nM, P < 0.001) compared to GBS patients. The support vector machine model incorporating BA profiles demonstrated strong discriminatory power (AUROC: 0.878), while a simplified tree-based algorithm using four key features achieved better performance (AUROC: 0.929).