The most studied area in dementia has been the hippocampus, with a well-demonstrated relationship between hippocampal volume and severity of disease and volume reduction over time. In addition, postmortem studies in AD have suggested that some of the earliest neuropathological changes, at least in terms of neurofibrillary tangles, start in the entorhinal cortex, and continue to the hippocampus before progressing to the association cortices. These findings accord with the considerable evidence that verbal and visual episodic memory dysfunction are typically the earliest symptoms of the disease and may be produced by left and right hippocampal damage. MRI volume analysis of temporal lobe structures in early disease confirm that progressive hippocampal atrophy occurs early in the disease. For a clinical trial in which hippocampal volume is to serve as a surrogate marker, several factors influence the choice of measurements to make: The measurements should be sensitive to the changes that occur in the earliest stages of AD. Jack et al. showed that hippocampal volumes in mild AD (CDR=0.5, MMSE=22) were 1.75 SDs below the control mean, while more severely affected AD subjects with a CDR of 1.0 and MMSE of 18 had a mean hippocampal volume 1.99 SDs below the control mean, in effect similar to the most mildly affected subjects. These results suggest the difference in hippocampal volume between AD subjects with a CDR of 0.5 and those with a CDR of 1.0 is only 2 to 3 percent of hippocampal volume. For the hippocampus, repeat volume measurements typically have an error margin of around 5 percent, depending on technique and experience, with amygdala measurement reproducibility being worse. Hippocampal volume losses in early AD have been estimated as ranging from 2 to 8 percent per year, with controls losing 0.7 percent/year and 2.3 percent per year in initially non-demented elderly subjects (>84 years) who went on to develop AD. Given the fact that the error margin is in the magnitude of the effect size large sample sizes will be needed. Other developments that are more promising in this respect are Voxel Based Morphometry, Registration and perhaps functional MRI. The first two detect whole brain volume changes that are probably more sensitive to change over time in AD, or in the transition between MCI and AD. Differences in volume are 3% per year in the AD population vs 0.2 % in controls with very small error margins. Functional MRI may be used to detect changes in activation patterns in reponse to therapy. However, given the limited availability of fMRI and the limited experience in patients this technique will only be usable in specific centres in subsamples.
In the context of a multicenter clinical trial one may expect greater variability, and therefore the ability to detect changes over time will require fairly large sample sizes and very careful quality control of scan acquisition and image analysis.
Back to Neuropsychology, brain aging and imaging
Back to Symposia
Back to The IPA European Regional Meeting (1-4 April 2003) of IPA