Center of Neuroimmunology Department of Neurology, Hospital Clinic of Barcelona, Institut d'Investigacions Biomèdiques August Pi Sunyer (IDIBAPS), University of Barcelona, Barcelona, Spain.
Department of Biomedical Engineering, Lerner Research Institute, Cleveland Clinic, Cleveland, Ohio.
JAMA Neurol. 2018 Oct 1;75(10):1246-1255. doi: 10.1001/jamaneurol.2018.1596.
Before using brain volume loss (BVL) as a marker of therapeutic response in multiple sclerosis (MS), certain biological and methodological issues must be clarified.
To assess the dynamics of BVL as MS progresses and to evaluate the repeatability and exchangeability of BVL estimates with Jacobian Integration (JI) and Functional Magnetic Resonance Imaging of the Brain (FMRIB) Software Library (FSL) (specifically, the Structural Image Evaluation, Using Normalisation, of Atrophy-Cross-Sectional [SIENA-X] tool or FMRIB's Integrated Registration and Segmentation Tool [FIRST]).
DESIGN, SETTING, AND PARTICIPANTS: A cohort of patients who had either clinically isolated syndrome or MS was enrolled from February 2011 through October 2015. All underwent a series of annual magnetic resonance imaging (MRI) scans. Images from 2 cohorts of healthy volunteers were used to evaluate short-term repeatability of the MRI measurements (n = 34) and annual BVL (n = 20). Data analysis occurred from January to May 2017.
The goodness of fit of different models to the dynamics of BVL throughout the MS disease course was assessed. The short-term test-retest error was used as a measure of JI and FSL repeatability. The correlations (R2) of the changes quantified in the brain using JI and FSL, together with the accuracy of the annual BVL cutoffs to discriminate patients with MS from healthy volunteers, were used to measure compatibility of imaging methods.
A total of 140 patients with clinically isolated syndrome or MS were enrolled, including 95 women (67.9%); the group had a median (interquartile range) age of 40.7 (33.6-48.1) years. Patients underwent 4 MRI scans with a median (interquartile range) interscan period of 364 (351-379) days. The 34 healthy volunteers (of whom 18 [53%] were women; median [IQR] age, 33.5 [26.2-42.5] years) and 20 healthy volunteers (of whom 10 [50%] were women; median [IQR] age, 33.0 [28.7-39.2] years) underwent 2 MRI scans within a median (IQR) of 24.5 (0.0-74.5) days and 384.5 (366.3-407.8) days for the short-term and long-term MRI follow-up, respectively. The BVL rates were higher in the first 5 years after MS onset (R2 = 0.65 for whole-brain volume change and R2 = 0.52 for gray matter volume change) with a direct association with steroids (β = 0.280; P = .02) and an inverse association with age at MS onset, particularly in the first 5 years (β = 0.015; P = .047). The reproducibility of FSL (SIENA) and JI was similar for whole-brain volume loss, while JI gave more precise, less biased estimates for specific brain regions than FSL (SIENA-X and FIRST). The correlation between whole-brain volume loss using JI and FSL was high (R2 = 0.92), but the same correlations were poor for specific brain regions. The area under curve of the whole-brain volume change to discriminate between patients with MS and healthy volunteers was similar, although the thresholds and accuracy index were distinct for JI and FSL.
The proposed BVL threshold of less than 0.4% per year as a marker of therapeutic efficiency should be reconsidered because of the different dynamics of BVL as MS progresses and because of the limited reproducibility and variability of estimates using different imaging methods.
在将脑容量损失（BVL）用作多发性硬化症（MS）治疗反应的标志物之前，必须明确某些生物学和方法学问题。
评估随着 MS 进展，BVL 的动态变化，并评估使用雅可比积分（JI）和大脑功能磁共振成像（FMRIB）软件库（FSL）（具体来说，结构图像评估，使用正常化，萎缩的矢状面[SIENA-X]工具或 FMRIB 的综合注册和分割工具[FIRST]）估计 BVL 的可重复性和可交换性。
设计、设置和参与者：从 2011 年 2 月到 2015 年 10 月，招募了一组患有临床孤立综合征或 MS 的患者。所有患者均接受了一系列年度磁共振成像（MRI）扫描。使用来自 2 组健康志愿者的图像来评估 MRI 测量的短期复测误差（n=34）和年度 BVL（n=20）。数据分析于 2017 年 1 月至 5 月进行。
评估了不同模型对 MS 病程中 BVL 动态的拟合程度。短期测试-再测试误差被用作 JI 和 FSL 可重复性的度量。使用 JI 和 FSL 量化的大脑变化的相关性（R2），以及每年 BVL 截止值来区分 MS 患者和健康志愿者的准确性，用于测量成像方法的兼容性。
共纳入 140 名患有临床孤立综合征或 MS 的患者，包括 95 名女性（67.9%）；该组的中位（四分位间距）年龄为 40.7（33.6-48.1）岁。患者接受了 4 次 MRI 扫描，中位（四分位间距）扫描间隔为 364（351-379）天。34 名健康志愿者（其中 18 名[53%]为女性；中位[IQR]年龄，33.5[26.2-42.5]岁）和 20 名健康志愿者（其中 10 名[50%]为女性；中位[IQR]年龄，33.0[28.7-39.2]岁）在中位（IQR）24.5（0.0-74.5）天和 384.5（366.3-407.8）天内分别进行了 2 次 MRI 扫描，用于短期和长期 MRI 随访。在 MS 发病后的前 5 年内，BVL 率较高（全脑容积变化的 R2为 0.65，灰质容积变化的 R2为 0.52），与类固醇呈直接相关（β=0.280；P=0.02），与发病年龄呈负相关，尤其是在发病后的前 5 年内（β=0.015；P=0.047）。FSL（SIENA）和 JI 的重复性对于全脑容积损失相似，而 JI 对特定脑区的估计更精确、偏差更小，而 FSL（SIENA-X 和 FIRST）则不然。使用 JI 和 FSL 测量的全脑容积损失之间的相关性很高（R2=0.92），但对于特定脑区的相关性较差。区分 MS 患者和健康志愿者的全脑容积变化的曲线下面积相似，尽管 JI 和 FSL 的阈值和准确性指数不同。
由于 MS 进展过程中 BVL 的动态变化以及使用不同成像方法估计的可重复性和可变性有限，建议重新考虑将每年 BVL 减少 0.4%作为治疗效果的标志物。
