Post-Mortem MR Brain Imaging Comparison with Macro- and Histopathology: Useful, Important and Underused
J.M. Wardlaw
2011
Cerebrovascular Diseases
Fax +41 61 306 12 34 E-Mail karger@karger.ch www.karger.com Comment Post-mortem MR is not straightforward. For example, air bubbles trapped in cortical sulci and in large basal arteries can create strong artifacts on some sensitive sequences, obscuring the underlying tissue [6] -although this is less of a problem with T 1 -and T 2 -weighted sequences. In general, although post-mortem and post-fixation tissue relaxation values may change, the differences in signal intensity between post-mortem
more »
... ssues are broadly similar to those seen in life [6] , enabling differentiation of normal and pathological structures, and the technique should be used much more. Post-mortem MR is both a highly valuable adjunct to routine post-mortem examination and as a research tool [3] . Some abnormalities are very obvious on MR, but less easily seen on macroscopic examination (like leukoaraiosis, small infarcts) [4] , so MR can direct the pathologist's attention to particular areas of interest. Brains can be imaged 'dry', or immersed in saline/formalin or embedded in gelatin. While testing ways of preparing the slices for imaging, De Reuck et al. [5] found that on images that were imaged 'dry', the BMBs and minibleeds appeared 5 times larger, and on slices immersed in saline they appeared 2 times larger, than their actual size at pathological examination. Others have demonstrated more detail of microvessels in vivo and in unembalmed brains than in formalin fixed brains at 8 T [7] , indicating the need for caution and for tailoring the examination to the specific question when setting up any post-mortem MR protocol -in the case of small vessel disease, one protocol does not fit all. The paper by De Reuck et al. [5] is only the fifth ever report of the histopathology of BMBs identified on imaging (sixth if a report on microbleeds in CADASIL is counted) and adds a further 20 patients to the previous literature total of 31, thereby nearly doubling the world literature on MR post-mortem comparisons of BMBs by taking the current total to 51. They studied haematomas, BMBs (defined as 1-3 mm diameter on macroscopic brain sections) and introduced a further small lesion, the brain minibleed, to describe lesions that were 0.2-0.5 mm diameter surrounding small vessels and not visible to the naked eye. They differentiated these from basal ganglia perivascular iron deposits [8, 9] . They determined the accuracy of T 2 * imaging at 7 T according to lesion size and location. Some important points emerge for the interpretation of imaging BMBs in life. Namely, all 26 haematomas, 76/79 BMBs and 138/163 minibleeds were correctly detected on T 2 * imaging; 25 lesions (mostly minibleeds and mostly in the deep white matter) were not detected on MR, and 41 signals that did not correspond to micro-or minibleeds were incorrectly detected with T 2 * -the latter were in fact perforating vessels containing post-mortem thrombi or perivascular iron deposits in the inferior basal ganglia not related to haemorrhage. Many imaging studies have tried to correlate BMB location with the likely underlying pathology and future risk of stroke or dementia. Lobar BMBs, i.e. at the cortico-subcortical junction, are said to be a marker of cerebral amyloid angiopathy, whereas 'Cerebral small vessel disease' is a rather broad term that describes a cluster of inter-related clinical and imaging features, including lacunar infarcts and haemorrhages, white matter lesions, enlarged perivascular spaces and microbleeds. These features often develop insidiously and, except for some lacunar stroke lesions, often without causing overt symptoms or signs until advanced. However, they are readily detected on magnetic resonance (MR) imaging, they increase in prevalence with advancing age, and their cause has been the subject of much speculation for many years. Indeed, there may be several causes. One source of speculation is the shortage of pathological studies of patients with any type of small vessel disease. Much of our understanding is based on work from the 1950s and 1960s, from the approximately 20 patients examined in detail by Miller Fisher [1] , prior to the advent of CT and MR imaging and the rapid subsequent decline in post-mortems. In addition to the lack of detailed pathological studies, there are few studies that compare specific lesions seen on imaging with their macroscopic pathological and histopathological counterparts [2] . The insidious onset, lack of discrete clinical expression, low mortality of lacunar stroke, and small size of the lesions all contribute to the shortage of detailed pathology early in disease, contrasting strongly with the abundance of imaging studies. Indeed some features (e.g. diffuse white matter lesions) are much more obvious on imaging than they are on macroscopic pathology [3, 4] , adding to the lack of detailed understanding of the corresponding pathology. Although imaging is now very sensitive, there are many aspects of the pathogenesis of small vessel disease that only detailed studies at a cellular level can tell us, thereby avoiding overzealous speculation about likely causes. Hence, the paper by De Reuck et al. [5] is a welcome addition to the scant literature [2] on pathological correlations of imagingdetected small vessel disease. They studied macro and microscopic features of brain microbleeds (BMBs) identified on formalinfixed post-mortem brain slices from patients with Alzheimer's disease using a 7-T small-bore MR scanner and T 2 * -weighted imaging. Their purpose was twofold: to establish a method for future imaging-pathology correlation studies and to examine the accuracy of MR identification of microbleeds compared with pathological examination.
doi:10.1159/000326846
pmid:21422756
fatcat:pwokzpalu5antn5te3jhzjp5fu