Measurement of brain atrophy in multiple sclerosis – a cross-sectional retrospective study

Background: Brain atrophy is a key feature of multiple sclerosis. To quantify brain atrophy with MRI, simple 2D measures like third ventricular width (TVW) and bicaudatio ratio (BCR) or computer-based segmentation procedures can be applied. The latter usually are generated using T1-weighted 3D data sets (MPRage sequences) and allow the quantification of the brain parenchymal fraction (BPF). Due to high methodological requirements, segmentation based methods are not part of the clinical routine yet.

Objective: To investigate in a cross-sectional setting the correlation between different measures of cerebral atrophy obtained from routine brain MRIs and clinical parameters.

Methods: All patients that had received a clinical evaluation and a brain MRI in our MS clinic from 1999 to 2002 and a clinical follow-up 2 years later were identified (n=82; relapsing MS=58, progressive MS=24; mean EDSS=3.7). Standardized clinical parameters including disease duration and EDSS were retrieved from the patient files. TVW and BCR were measured manually from axial T1- and FLAIR-weighted images. BPF was calculated from axial T2-weighted scans (n=57) with SPM2 automatically or following manual correction from 2 independent examiners.

Results: We found significant correlations between manually measured 2D parameters (TVW/BCR) with age (0.43; p<0.001/0.50; p<0.001), EDSS (0.44; p<0.001/0.36; p<0.001), and disease duration (0.41; p<0.001/0.47; p<0.001). BPF measured by segmentation of T2-weighted images significantly correlated with age (-0.46; p<0.001), EDSS (0.45; p<0.001), walking distance (0.44; p<0.001), TVW (-0.49; p<0.001), and BCR (-0.43; p<0.005). For the subgroup of RRMS patients, weaker correlations were found for age (-0.34; p<0.05) and TVW (-0.33; p<0.05). Strong correlations were found in SPMS and PPMS patients, in particular for EDSS (-0.59; p<0.05), walking distance (0.69; p<0.005), TVW (-0.69; p<0.005), and BCR (-0.66; p<0.005).

Conclusions: The measurement of atrophy parameters obtained from routine brain MRIs manually and after segmentation of T2-weighted 2D data sets is feasible and correlates with clinical parameters. From the literature similar correlation coefficients using T1-based segmentations are known. An accompanying study describes the predictive value of the investigated atrophy parameters for disease progression in MS.