Contributions
Abstract: P1037
Type: Poster
Abstract Category: Pathology and pathogenesis of MS - Imaging
Background: Brain volume loss (BVL) on MRI is among the best imaging correlates of MS disease progression. However, measures of BVL can require hundreds of participants and long follow-up to see changes due to treatment. Some tissue pathology, e.g. demyelination and axonal loss, with a resulting increase in extracellular space, represent tissue loss that is not fully captured by BVL measurements, but, rather, decrease the density (and alter signal intensity) of remaining brain tissue. Other factors, e.g. edema and hydration, can spuriously affect BVL. We propose a method to combine measurements of tissue density and volume into “signal mass,” (SM) a more comprehensive measure of tissue loss and atrophy in MS.
Objectives: Compare SM and Jacobian Integration (JI) for detecting overall tissue loss (atrophy) in MS versus healthy controls.
Methods: 9 MS and 9 healthy controls had T1-weighted (T1w) and magnetization transfer ratio (MTR) MRI at baseline and two years, at 1.5 and 3T. MTR was used for SM because it is sensitive to changes in tissue integrity and is consistent over time on the same scanner. T1w images were co-registered using a rigid skull-constrained transform to a half-way space, followed by nonlinear registration. MTR images were registered to the T1w images, then transformed using the T1w transforms. Parenchyma masks were made using tissue maps from SIENAX. Jacobian determinant maps (J) were computed from the nonlinear transforms. The percent brain volume change (PBVC) was the voxel-wise average of J within the parenchyma mask. MTR SM at baseline was S1 = I1 ´ vol where I1 is the baseline MTR image and vol is the voxel volume. At followup this was S2 = I2 ´ vol ´ J. Percent SM change (PSMC) was the percent difference between S1 and S2. JI and SM were both annualized. Required N for one-tailed t-tests comparing healthy control to MS were computed for each measurement using a = 0.05 and b = 0.9, at each field strength.
Results: At 1.5 T PBVC loss was 0.13 %/year greater in patients than controls, s = 0.34; PSMC loss was 0.33 %/year greater, s = 0.56. The required N per arm to detect these differences are 108 for JI and 45 for SM. A trial expecting a 50% treatment effect would require N of 430 for JI and 181 for SM. These results were confirmed in the 3 T data.
Conclusions: Signal mass demonstrated a better than 2x improvement in N over JI for detecting a difference from normal, and may be a more comprehensive measurement of tissue loss in MS.
Disclosure:
This research was funded by a grant from the Canadian Institutes of Health Research.
Dr. Brown has received personal compensation for consulting services from NeuroRx Research.
Dr. Narayanan has received personal compensation from NeuroRx Research for consulting activities, and a speaker"s honorarium from Novartis Canada.
Dr. Arnold reports personal fees for consulting from Acorda, Biogen, Hoffman LaRoche, MedImmune, Mitsubishi, Novartis, Receptos, Sanofi-Aventis; grants from Biogen and Novartis; and an equity interest in NeuroRx Research.
Abstract: P1037
Type: Poster
Abstract Category: Pathology and pathogenesis of MS - Imaging
Background: Brain volume loss (BVL) on MRI is among the best imaging correlates of MS disease progression. However, measures of BVL can require hundreds of participants and long follow-up to see changes due to treatment. Some tissue pathology, e.g. demyelination and axonal loss, with a resulting increase in extracellular space, represent tissue loss that is not fully captured by BVL measurements, but, rather, decrease the density (and alter signal intensity) of remaining brain tissue. Other factors, e.g. edema and hydration, can spuriously affect BVL. We propose a method to combine measurements of tissue density and volume into “signal mass,” (SM) a more comprehensive measure of tissue loss and atrophy in MS.
Objectives: Compare SM and Jacobian Integration (JI) for detecting overall tissue loss (atrophy) in MS versus healthy controls.
Methods: 9 MS and 9 healthy controls had T1-weighted (T1w) and magnetization transfer ratio (MTR) MRI at baseline and two years, at 1.5 and 3T. MTR was used for SM because it is sensitive to changes in tissue integrity and is consistent over time on the same scanner. T1w images were co-registered using a rigid skull-constrained transform to a half-way space, followed by nonlinear registration. MTR images were registered to the T1w images, then transformed using the T1w transforms. Parenchyma masks were made using tissue maps from SIENAX. Jacobian determinant maps (J) were computed from the nonlinear transforms. The percent brain volume change (PBVC) was the voxel-wise average of J within the parenchyma mask. MTR SM at baseline was S1 = I1 ´ vol where I1 is the baseline MTR image and vol is the voxel volume. At followup this was S2 = I2 ´ vol ´ J. Percent SM change (PSMC) was the percent difference between S1 and S2. JI and SM were both annualized. Required N for one-tailed t-tests comparing healthy control to MS were computed for each measurement using a = 0.05 and b = 0.9, at each field strength.
Results: At 1.5 T PBVC loss was 0.13 %/year greater in patients than controls, s = 0.34; PSMC loss was 0.33 %/year greater, s = 0.56. The required N per arm to detect these differences are 108 for JI and 45 for SM. A trial expecting a 50% treatment effect would require N of 430 for JI and 181 for SM. These results were confirmed in the 3 T data.
Conclusions: Signal mass demonstrated a better than 2x improvement in N over JI for detecting a difference from normal, and may be a more comprehensive measurement of tissue loss in MS.
Disclosure:
This research was funded by a grant from the Canadian Institutes of Health Research.
Dr. Brown has received personal compensation for consulting services from NeuroRx Research.
Dr. Narayanan has received personal compensation from NeuroRx Research for consulting activities, and a speaker"s honorarium from Novartis Canada.
Dr. Arnold reports personal fees for consulting from Acorda, Biogen, Hoffman LaRoche, MedImmune, Mitsubishi, Novartis, Receptos, Sanofi-Aventis; grants from Biogen and Novartis; and an equity interest in NeuroRx Research.