There was a lot of talk of images at this meeting, most of it cordial. One session included a number of presentations addressing aspects of the use of various imaging modalities to diagnose and/or follow the course of AD. For example, Jonathan Chalk reported on a 12-month longitudinal study of AD progression using MRI. The rate of lateral ventricular atrophy was not as robust as hoped for. (Unfortunately, his data on this point didn’t show up on the slides, just the titles. Rather ironic for a talk on imaging!) Temporal lobe atrophy rate also showed overlap between AD and controls but the atrophy rates for the cerebrum in AD was about six times greater than in controls. Lateral ventricle rates were 7 times greater and temporal lobe atrophy was 5 times greater. Perhaps the most interesting aspect of this talk was the calculations of sample size needed to detect a treatment effect of 20 percent (85 patients to detect lateral ventricular atrophy, 189 for global atrophy and 288 for temporal lobe atrophy). Chalk also noted that a persistent technical limitation is the head movement of patients who are in the scanner.

Rachael Scahill followed with an intriguing display of images based on a method known as fluid-registered serial MRI. A comparison of images taken from the same patient over a period of several months will demonstrate areas of tissue loss and tissue gain. But it doesn’t actually reveal where the tissue is lost because it is based on comparisons of boundaries. To localize tissue loss, a different model is required in which the second scan is “warped” (are you listening Star Trek fans?) onto the first scan to provide a match that reveals measures of contraction or expansion. With this model it is clear that the atrophy is primarily in the temporal lobe (not surprisingly). When similar comparisons are made across several patients it is then possible to “normalize” the results using a template image in known standard space. What is seen with this kind of comparison? In mild cases, there is an expansion of the ventricles but contraction of posterior cingulate and temporal lobe regions (consistent with PET data). In later cases there appears to be a shift in atrophy from medial to lateral temporal lobe regions although it is not clear if this is an artifact.

Eric. Reiman reported on fluoro-DG-PET to track progression of cognitive changes in ApoE4 patients prior to any evidence of disease. They had earlier shown decreased glucose metabolism in brain regions that ultimately are affected by the disease in middle-aged patients but had recently extended the analysis to young (20-39-year old) subjects: 15 controls and 12 E3/E4 subjects who were matched for age, gender, education, MMSE and gender. No differences in neuropsychological performance were apparent. No differences in whole brain or pontine glucose metabolism were found, but parietal, temporal and posterior cingulate regions showed bilaterally reductions of around 10 percent in AD patients. These changes apparently precede the appearance of plaques and tangles with the exception of tangles in the entorhinal cortex. If follow-up indicates that these metabolic changes are predictive of disease risk, they may be used to monitor primary prevention therapies.

The most spirited discussion followed a talk by Daniel Silverman, who addressed the role of PET as a diagnostic tool. Based on the number of questions raised by this talk, I think it is fair to say that there is no consensus on the utility of PET for diagnostic purposes, but there was clearly a lot of interest in the extent to which imaging matches up with clinical and postmortem assessments. Other talks in this session supported the continuing role of imaging methods as both a diagnostic and experimental tool. One can only hope that the ability to monitor the progression of the disease will coincide with the development of therapies that interfere with its insidious course.