This paper is a very well performed study, extremely interesting, and has potentially great significance for the future. The researchers are using high magnetic fields, but human MRI systems with these magnetic fields are now available for research and in the future may have more general applicability. Someday, this could make its way into a hospital setting. This is a different approach from current MRI, with or without contrast agents. There is considerable interest in various techniques using MRI and PET to detect amyloid and other proteins associated with neurodegeneration, and this excellent paper represents one approach.

View all comments by Michael Weiner

Many researchers and clinicians agree that there is great value in developing imaging technologies for the quantitation of amyloid deposition in human brain. Potential benefits of this effort include improved diagnosis of AD, particularly in early and confusing cases of dementia, and the prospect of presymptomatic identification of AD pathology. The latter possibility becomes critical, if it proves necessary to begin anti-amyloid therapies as early in the course of amyloid deposition as possible and if waiting for even mild clinical symptoms (e.g., MCI) proves to be waiting too long. What’s more, imaging amyloid in vivo is likely to be important to the development of these anti-amyloid therapies, since information about drug efficacy on the primary target would seem to be invaluable in the development of any drug.

Several groups have mounted efforts towards realization of this goal. Our group in Pittsburgh, (Pittsburgh Compound-B) along with the groups at UCLA (FDDNP), the University of Pennsylvania (IMPY), the University of Toronto (SB-13), and the BF Research Institute in...  Read more

Many researchers and clinicians agree that there is great value in developing imaging technologies for the quantitation of amyloid deposition in human brain. Potential benefits of this effort include improved diagnosis of AD, particularly in early and confusing cases of dementia, and the prospect of presymptomatic identification of AD pathology. The latter possibility becomes critical, if it proves necessary to begin anti-amyloid therapies as early in the course of amyloid deposition as possible and if waiting for even mild clinical symptoms (e.g., MCI) proves to be waiting too long. What’s more, imaging amyloid in vivo is likely to be important to the development of these anti-amyloid therapies, since information about drug efficacy on the primary target would seem to be invaluable in the development of any drug.

Several groups have mounted efforts towards realization of this goal. Our group in Pittsburgh, (Pittsburgh Compound-B) along with the groups at UCLA (FDDNP), the University of Pennsylvania (IMPY), the University of Toronto (SB-13), and the BF Research Institute in Osaka, Japan (BF-168) have all developed amyloid-binding agents designed to be tracers for either PET or SPECT imaging. Several of these compounds have been used in human studies. [F-18]FDDNP is being used in human studies at UCLA. Pittsburgh Compound-B was first used in humans in Uppsala, Sweden, was subsequently used in human studies in Pittsburgh, and is currently being actively used in eight PET centers worldwide (with preparations being made for use in several more). SB-13 has been applied to human studies in Toronto in collaboration with the Penn group. All of these agents have been reported to image cortical amyloid deposits in AD subjects. The approach in these studies is analogous to the approach used for decades in neuroreceptor imaging studies. Other than the considerable radiochemical effort required to develop the amyloid probes, no new technology was required to apply these probes to human studies allowing this approach to be rapidly translated to human studies.

Even with the recent conditional CMS approval of FDG PET scanning for the diagnosis of AD, PET scanning remains less accessible and more expensive than MRI. Although PET imaging is readily quantifiable, a disadvantage is the relatively low resolution of PET. In addition, microPET studies in transgenic mice with the same ligands that have been successful in humans have not been successful to date. Therefore, development of an MRI-based technology for amyloid imaging in transgenic mice and/or humans would be an important addition to the imaging arsenal in AD research. The study by Jack et al. represents an important step in that direction. Two approaches have been taken in MR efforts to image amyloid. One is similar to the PET/SPECT approach in that it seeks to employ an amyloid-binding MR contrast agent detectable by MRI. Thus far, these amyloid-binding MR contrast agents have been modified fragments of Aβ itself labeled either with gadolinium or microcrystalline iron oxide nanoparticles (MION). This approach differs importantly from the PET/SPECT approach in that no precedent exists for labeling brain parenchymal targets with these types of contrast agents. In addition, macromolecules as large as Aβ have not yet been successfully used as imaging agents. Thus, a major technological hurdle needs to be crossed before the application of this MRI approach becomes a reality in human studies. Gadolinium- and MION-labeled Aβ derivatives have been used to label plaques in transgenic mice in vivo, but the required Aβ concentrations, long imaging times, or need for mannitol to open the blood-brain barrier makes application to human studies difficult.

This MRI study, by Jack et al., differs from previous MRI studies in important ways. First, no contrast agent is used. Plaques are imaged based on inherent physicochemical differences from surrounding tissue. Thus, no unique hurdle stands in the way of application to human studies. Second, while plaques have previously been imaged without contrast agents in postmortem human and mouse tissue, those studies typically used very long imaging times (up to 24 hours) to produce useful images. Jack et al. were able to obtain useful images in 60-90 minutes, times potentially applicable to human studies. The high resolution of MRI proves to be both a strength and a weakness of this approach. It is a strength because, unlike PET which blurs the ~50 mm individual plaques into regional average plaque load of 0.5-1 cm resolution, MRI can resolve individual plaques. This can also be a disadvantage because MRI must be performed in a way to image individual plaques because “blurring” would cause the experiment to degenerate into measurements of tissue T2 differences which has not proven to be successful in AD studies. Therefore, the limiting factor in these in vivo MRI studies proves to be motion artifacts because movements of less than 0.1 mm may be sufficient to blur the plaques. This means that inherent tissue movement due to pulsations in blood pressure caused by the beating heart and movements associated with respiration become significant. Jack et al. elegantly overcome this problem in the present study by cardiorespiratory triggering of the pulse sequence during free breathing. Of course, gross movements needed to be controlled as well by fixation of the head in a rigid device. Use of a specialized long TE spin echo pulse sequence at 9.4T further minimized sensitivity to motion-related blurring.

The identity of plaques imaged in vivo was convincingly demonstrated by subsequent ex vivo MRI and standard histology of brain sections. However, it appears that only a subset, perhaps only 10-15 percent, of plaques can be visualized in vivo and it is not clear what determines which plaques can be detected, although size appears to be the most important factor. Another question is whether this technique can be quantified beyond simply counting plaques or recording total volume of dark spots. The latter would be similar to semi-quantitative measurements of percent area occupied by immunohistologically stained plaques on tissue sections and may be sufficient. Thus, this technique already appears to be a viable tool for imaging the natural history of individual plaque development in transgenic mice and detailed effects of anti-amyloid therapies at the plaque level. That alone represents an important advance, given the failure of microPET approaches to imaging amyloid in mice. Jack et al. acknowledge that this technology requires improvements in instrumentation before application to human studies is feasible. To date, the use of a field strength as high as 9.4T in human studies has proven problematic. In addition, Jack et al. acknowledge that the aged human brain may have many “dark spot” artifacts not present in mouse brain such as microhemorrhages which could be mistaken for plaques. Perhaps the most difficult challenge will be limitation of movement artifacts in human studies to the submillimeter level. If this proves to be a tractable problem, MRI visualization of individual amyloid plaques could be an important complement to the more macroscopic measurement of plaque load currently provided by PET.

View all comments by William Klunk