Fortanier Etienne, Michel Constance P, Hostin Marc Adrien, Delmont Emilien, Verschueren Annie, Guye Maxime, Bellemare Marc-Emmanuel, Bendahan David, Attarian Shahram
Reference Center for Neuromuscular Diseases and ALS, La Timone University Hospital, Aix-Marseille University, Marseille, France.
Center for Magnetic Resonance in Biology and Medicine, Aix-Marseille University, UMR CNRS 7339, Marseille, France.
Eur J Neurol. 2025 Jan;32(1):e16574. doi: 10.1111/ene.16574. Epub 2024 Nov 27.
Intramuscular fat fraction (FF), assessed using quantitative MRI (qMRI), has emerged as a promising biomarker for hereditary transthyretin amyloidosis (ATTRv) patients. Currently, the main drawbacks to its use in future therapeutic trials are its sensitivity to change over a short period of time and the time-consuming manual segmentation step to extract quantitative data. This pilot study aimed to demonstrate the suitability of an Artificial Intelligence-based (AI) segmentation technique to assess disease progression in a real-life cohort of ATTRv patients over 1 year.
Fifteen ATTRv patients were included in this monocentric, observational, prospective study. FF, magnetization transfer ratio (MTR), and quantitative T2 were extracted from patients' lower limb qMRI at two time points, 1 year apart, at thigh and leg levels. qMRI parameters were correlated with clinical and electrophysiological parameters assessed at the same time.
Global FF at leg level significantly progressed over 1 year: +1.28 ± 2.62% (p = 0.017). At thigh level, no significant change in global FF, MTR, or T2 was measured. The leg FF was strongly correlated with the main clinical and electrophysiological scores.
AI-based CNN network segmentation combined with qMRI can be used to obtain quantitative metrics for longitudinal studies in ATTRv patients. Global FF at the leg level seems to be the most sensitive MRI biomarker to track disease progression in a 1-year period. Larger studies with treatment-specific groups will now be necessary to determine the place of qMRI markers compared to the current clinical and electrophysiological scores.
使用定量磁共振成像（qMRI）评估的肌肉内脂肪分数（FF）已成为遗传性转甲状腺素蛋白淀粉样变性（ATTRv）患者一种有前景的生物标志物。目前，在未来治疗试验中使用它的主要缺点是其在短时间内对变化的敏感性以及提取定量数据时耗时的手动分割步骤。这项初步研究旨在证明基于人工智能（AI）的分割技术在评估ATTRv患者真实队列1年期间疾病进展方面的适用性。
15例ATTRv患者纳入了这项单中心、观察性、前瞻性研究。在相隔1年的两个时间点，从患者大腿和小腿水平的下肢qMRI中提取FF、磁化传递率（MTR）和定量T2。qMRI参数与同时评估的临床和电生理参数相关。
小腿水平的整体FF在1年期间显著进展：增加1.28±2.62%（p = 0.017）。在大腿水平，未测得整体FF、MTR或T2有显著变化。小腿FF与主要临床和电生理评分密切相关。
基于AI的卷积神经网络（CNN）分割结合qMRI可用于获取ATTRv患者纵向研究的定量指标。小腿水平的整体FF似乎是在1年期间追踪疾病进展最敏感的MRI生物标志物。现在需要开展针对特定治疗组的更大规模研究，以确定与当前临床和电生理评分相比qMRI标志物的地位。
