MULTIPLE SCLEROSIS MSJ JOURNALCerebral atrophy has been used for centuries by pathologists as a marker of the loss of cerebral tissue due to injury or degeneration, and the tissue damage and loss associated with atrophy has been well documented. Thus, atrophy as a marker of neurodegeneration has incontrovertible face validity, at least over the long term. So, why is the use of whole brain atrophy as an outcome measure in multiple sclerosis (MS) the subject of debate?Magnetic resonance imaging (MRI)-based computational methods provide sensitive and reproducible measures of brain volumes that are capable of detecting changes on the order of 0.2%: changes that are much smaller than those described at the time of postmortem examination. The interpretation of such small changes, particularly over the short term, involves subtleties that are not relevant over the long term. For example, small physiological fluctuations, due to shifts in water, 1 can produce brain volume fluctuations of this order. Furthermore, the accelerated atrophy seen after initiation of some anti-inflammatory therapies, often referred to as pseudoatrophy, also complicates the interpretation of changes in brain volume. Although the pathophysiological mechanisms responsible for pseudoatrophy are still unclear, pseudoatrophy appears to be related to the nature of the anti-inflammatory therapy and the amount of inflammation present at the start of therapy.Despite these limitations, as argued by Rudick and Fisher, a substantial body of evidence supports the use of whole brain atrophy, at least after the first year following initiation of anti-inflammatory therapy, as a marker of the end-stage of tissue injury of whatever sort and, as such, as a candidate for the best MRI marker of effective therapy in MS.Filippi and Rocca also value atrophy as an outcome measure but, because it is an end-stage process, argue for outcome measures that provide a more pathologically specific indication of the processes that contribute to tissue loss, including gray matter (GM) atrophy, magnetization transfer imaging (MTI), diffusion tensor imaging (DTI), and magnetic resonance spectroscopy (MRS). Each of these measures has advantages and disadvantages.Since GM is less inflamed than white matter (WM) in MS, it should be less susceptible to pseudoatrophy. However, the correct segmentation of GM on MRI is challenging 2 and less precise than the segmentation of whole brain. Changes in magnetization transfer have relative specificity for changes in myelin density, which is of particular interest in MS. However, the changes that occur in whole brain magnetization transfer are extremely small 3 and, as in the case of whole brain atrophy, the interpretation of such small changes may involve subtleties related not only to remyelination but also to other physiological changes that affect myelin density, such as changes in brain water or the relative partial volume of other cell types. DTI seems to be less pathologically specific for axons or myelin than one would like. MRS is patho...