Contributions
Abstract: P583
Type: Poster
Abstract Category: Pathology and pathogenesis of MS - 21 Imaging
Objective: To determine and compare various sources of variability in brain volume (BV) measurements in MS.
Background: BV measurements fluctuate, even when using standardized MRIs and precise measurement techniques, complicating interpretation of changes in individuals. The degree to which various sources of variability contribute to fluctuations in BV has not been extensively studied.
Methods: MS patients were enrolled in a scan-rescan study at 3 MS PATHS participating healthcare institutions. Each patient had 2 visits within 7 days, with 2 MRIs at each visit for a total of 4 MRIs. MRIs were acquired on 2 different Siemens 3T scanners for each patient using identical protocols that included 1 mm isotropic 3D sequences (FLAIR, MPRAGE). Half of the patients had MRIs done on the same scanner on the same day. BV was measured using fully-automated software to calculate brain parenchymal fraction (BPF) (autosegMS, Cleveland Clinic). Measurements performed on different scanners were analyzed before and after calibration. Measurement errors (intra- and inter-scanner variability) and physiologic (between-day) variability were expressed as relative standard deviations (RSD) estimated from a variance components model; 95% confidence intervals for the RSD were estimated via a bootstrap percentile method.
Results: Thirty patients participated: Expanded Disability Status Scale (EDSS) 0-6, age 23-55, disease duration 2-27 years. Scanner models: 1 Verio, 5 Skyras, 1 Prisma, and 1 MR/PET. In total, 120 images were analyzed. The components of variability in BPF were estimated as: 0.17% [0.13-0.21%] due to intra-scanner measurement error, 0.21% [0.10-0.29%] and 0.05% [0.00-0.14%] due to inter-scanner measurement error before and after scanner calibration, respectively, and 0.20% [0.09-0.28%] due to physiological variability. The worst-case variability estimated from scans acquired on a different day and different scanner was 0.34% [0.30-0.42%] before calibration and 0.25% [0.23-0.32%] after calibration.
Conclusion: The worst-case variability in BV can be reduced by approximately 25% using a simple linear calibration between scanners. Day-to-day physiologic variability was similar in magnitude to the inter-scanner measurement error. A better understanding of the sources of variability may lead to improved techniques for estimating the extent of true tissue loss and enable the use of BV measurements in routine clinical care of MS patients.
Disclosure: This study was sponsored by Biogen, Inc.
A. Tsang, C. Wager, D. Jennings, C. de Moor, R. Rudick, J.R. Williams and E. Fisher are employees of, and stockholders in, Biogen.
R. Bermel has received consulting fees from Biogen, Novartis, Genentech, Genzyme and Mallinckrodt, and is part contributor to the intellectual property used in the MSPT.
S.E. Jones has received a speaker fee from Monteris and Siemens.
I. Izbudak has nothing to disclose.
Y.W. Lui has nothing to disclose.
L. Krupp has received consultant fees from EMD Serono, Projects in Knowledge, Novartis, Pfizer (for serving on a DSMB), Biogen, PK Law and Teva Neurosciences; Royalty payments from Abbvie Inc. and Grifols World Wide Services; and research grant support from Novartis, Teva Neurosciences, Biogen, NIH, National Multiple Sclerosis Society, Department of Defense and the Lourie Foundation.
E.M. Mowry has received research funding from Biogen and free medication for a clinical trial from Teva Neuroscience. She is a site PI for trials funded by Biogen and Sun Pharma, and receives royalties for editorial duties with UpToDate.
Abstract: P583
Type: Poster
Abstract Category: Pathology and pathogenesis of MS - 21 Imaging
Objective: To determine and compare various sources of variability in brain volume (BV) measurements in MS.
Background: BV measurements fluctuate, even when using standardized MRIs and precise measurement techniques, complicating interpretation of changes in individuals. The degree to which various sources of variability contribute to fluctuations in BV has not been extensively studied.
Methods: MS patients were enrolled in a scan-rescan study at 3 MS PATHS participating healthcare institutions. Each patient had 2 visits within 7 days, with 2 MRIs at each visit for a total of 4 MRIs. MRIs were acquired on 2 different Siemens 3T scanners for each patient using identical protocols that included 1 mm isotropic 3D sequences (FLAIR, MPRAGE). Half of the patients had MRIs done on the same scanner on the same day. BV was measured using fully-automated software to calculate brain parenchymal fraction (BPF) (autosegMS, Cleveland Clinic). Measurements performed on different scanners were analyzed before and after calibration. Measurement errors (intra- and inter-scanner variability) and physiologic (between-day) variability were expressed as relative standard deviations (RSD) estimated from a variance components model; 95% confidence intervals for the RSD were estimated via a bootstrap percentile method.
Results: Thirty patients participated: Expanded Disability Status Scale (EDSS) 0-6, age 23-55, disease duration 2-27 years. Scanner models: 1 Verio, 5 Skyras, 1 Prisma, and 1 MR/PET. In total, 120 images were analyzed. The components of variability in BPF were estimated as: 0.17% [0.13-0.21%] due to intra-scanner measurement error, 0.21% [0.10-0.29%] and 0.05% [0.00-0.14%] due to inter-scanner measurement error before and after scanner calibration, respectively, and 0.20% [0.09-0.28%] due to physiological variability. The worst-case variability estimated from scans acquired on a different day and different scanner was 0.34% [0.30-0.42%] before calibration and 0.25% [0.23-0.32%] after calibration.
Conclusion: The worst-case variability in BV can be reduced by approximately 25% using a simple linear calibration between scanners. Day-to-day physiologic variability was similar in magnitude to the inter-scanner measurement error. A better understanding of the sources of variability may lead to improved techniques for estimating the extent of true tissue loss and enable the use of BV measurements in routine clinical care of MS patients.
Disclosure: This study was sponsored by Biogen, Inc.
A. Tsang, C. Wager, D. Jennings, C. de Moor, R. Rudick, J.R. Williams and E. Fisher are employees of, and stockholders in, Biogen.
R. Bermel has received consulting fees from Biogen, Novartis, Genentech, Genzyme and Mallinckrodt, and is part contributor to the intellectual property used in the MSPT.
S.E. Jones has received a speaker fee from Monteris and Siemens.
I. Izbudak has nothing to disclose.
Y.W. Lui has nothing to disclose.
L. Krupp has received consultant fees from EMD Serono, Projects in Knowledge, Novartis, Pfizer (for serving on a DSMB), Biogen, PK Law and Teva Neurosciences; Royalty payments from Abbvie Inc. and Grifols World Wide Services; and research grant support from Novartis, Teva Neurosciences, Biogen, NIH, National Multiple Sclerosis Society, Department of Defense and the Lourie Foundation.
E.M. Mowry has received research funding from Biogen and free medication for a clinical trial from Teva Neuroscience. She is a site PI for trials funded by Biogen and Sun Pharma, and receives royalties for editorial duties with UpToDate.