Amyloid deposits can be rapidly detected in the brains of living mice using a novel ligand and near-infrared fluorescence imaging.
Until recently using clinical imaging technologies such as positron emission tomography (PET) and magnetic resonance imaging (MRI), the amyloid plaques that accumulate in the brains of patients with Alzheimer disease have been difficult, if not impossible, to detect in vivo. In this issue Hintersteiner et al.1 describe a different approach to imaging amyloid. Using a near infrared (NIR) fluorescence probe that crosses the blood-brain barrier and binds amyloid plaques in the brains of mice, the amount of amyloid can be cost-effectively estimated using near infrared fluorescence imaging. Eventually such an approach may be adapted to visualize amyloid in humans.
The ability to image disease processes in living humans is one of the major technologic advances of modern medicine. In the context of disease management, in vivo imaging is one of the many, and often most informative, modalities that can be used to diagnose diseases and evaluate treatment outcomes. Largely because of costs, imaging is less commonly used to predict risk for the development of disease in asymptomatic individuals. Although not always thought of as such, tests that rely on imaging are fundamentally biomarker studies. As with any biomarker assay, the utility of such tests depends on the sensitivity and specificity of the biomarker and on the sensitivity and specificity of the test used to measure that biomarker. The former issue is extremely important to recognize. No matter how good the assay, its predictive ability is only as good as the predictive ability of the biomarker being studied.
Deposition of the amyloid β-peptide into a fibrillar β-sheet structure referred to as amyloid is a diagnostic hallmark of the post-mortem Alzheimer-disease brain. In Alzheimer disease, amyloid β deposits as amyloid in senile plaques and, more variably, in cerebral vessels. The amyloid in senile plaques forms a spherical core that ranges from ∼2 to ∼200 μm, but is typically 20–60 μm in diameter. Accumulation of amyloid β in the brain is also hypothesized to be the cause of Alzheimer disease, although there is a great deal of debate as to whether the toxic form is amyloid β that accumulates as visible amyloid deposits or as smaller soluble extracellular or intracellular aggregates. In any case, there is a great deal of evidence from genetic, biochemical, pathologic and animal modeling studies to support the hypothesis that accumulation of amyloid β in the brain is the initiating event in Alzheimer-disease pathogenesis2. Thus, in theory, amyloid β deposition as amyloid appears to be a good diagnostic biomarker for Alzheimer disease and may also be a good predictive biomarker of the disease.
Given the potential of amyloid as a biomarker, a number of groups have been developing approaches to visualize plaques. Conceptually, the simplest approach has been to use MRI to try to directly visualize amyloid plaques. Despite recent impressive technological advances in MRI resolution, this approach remains challenging because of the plaques' small size. Several recent reports suggest that this may now be feasible, at least in transgenic mouse models and in mouse and human tissue slices3,4. However, only large plaques (>50 μM) can currently be visualized, and the scans require hours to complete. Thus, additional advances in MRI technology would be needed to translate this approach to humans or even to make it more widely accessible for preclinical studies of amyloid β deposition in mice.
Compounds such as Congo red (Fig. 1a) and thioflavin S have been used for many years to recognize amyloid deposits in postmortem tissue. Thus, another approach to visualizing amyloid β deposited in vivo is to modify known amyloid-binding compounds or identify new ones. The most advanced of these ligand-based approaches uses Pittsburgh compound B (PIB) to visualize plaque burden in living humans with PET5. PIB is a thioflavin derivative labeled with 11C so that it can be used as a PET probe (Fig. 1a). PIB is the only amyloid probe currently used to image amyloid in Alzheimer disease patients, but because of its only recent approval for humans and somewhat limited availability, the clinical utility of PET imaging with PIB has not been definitively established.
Although PIB has been used to image amyloid plaque in humans in vivo and in mice tissue ex vivo, microPET with PIB has not been able to detect amyloid in the brains of living Alzheimer-disease mouse models. Moreover, the inherent resolution limits of PET impose certain restrictions on the information that this approach can provide. For example, it is unlikely that PET scans will ever be able to detect individual plaques or even to provide precise localization of amyloid deposition within structural subregions of the brain.
Other studies suggest that combining MRI with an amyloid ligand labeled with 1H or 19F may enable visualization of amyloid plaques6,7,8. The most recent report shows that MRI can detect an 19F-labeled amyloid dye known as FSB and that FSB labeling is superior to other MRI methods for detecting amyloid plaques in living mice9. Although such approaches provide more spatial resolution than PET studies and avoid the use of radioactive isotopes with short half-lives, they are still subject to some of the technical limitations of traditional MRI studies and, at least with current ligands, to suboptimal signal-to-noise levels.
Hintersteiner et al. describe an alternative, all-optical approach to imaging an amyloid ligand in vivo. Using a novel fluorescent amyloid-binding dye, AOI987 (Fig. 1a), which absorbs and emits in the NIR spectrum, they visualize amyloid load in the brain of living mice using NIR fluorescence imaging (Fig. 1b). Fluorescent imaging in live animals with shorter wavelengths of light is severely limited because of tissue autofluorescence and scattering. However, NIR imaging solves this problem by reducing the background and scattering through biological tissue10. Although the spatial resolution in the work of Hintersteiner et al. is limited, the authors do demonstrate that the fluorescence signal intensity increases with increasing plaque load in the mice and that amyloid deposition can be detected in mice as young as 9 months. Moreover images could be acquired in short time periods ranging from 0.5 to 3.0 seconds, allowing measurement at multiple time points after dosing to calculate specific binding.
Despite the lack of spatial resolution, the ability to noninvasively quantify amyloid β deposited as amyloid in living mice using NIR imaging is quite exciting. Unlike PET and MRI imaging technologies, which require significant hardware investments, highly skilled operators, and, in the case of PET, short-lived radioligands, in vivo NIR imaging as described in this study is quite inexpensive and could be set up in virtually any laboratory.
It should also be possible to improve the spatial resolution. NIR imaging using the more elaborate techniques and equipment of optical tomography (which was not done in this study) has attained a spatial resolution of 1 mm or better, approximately the resolution of PET imaging11. It may also be possible to improve upon the fluorescent signal by improving the amyloid probe. For example, an amyloid dye that fluoresced more intensely, changed spectra, or both, upon binding to amyloid could reduce background and improve contrast.
Currently, the major constraint on NIR fluorescence imaging is the paucity of fluorescent probes. Probes described in the literature are ligands tagged with NIR fluorophores. Thus, on a more general level, the finding of Hintersteiner et al. is rather exciting simply because they identify an NIR fluorescent ligand that binds the biomarker of interest. This proof-of-concept study will likely engage others in the search for better NIR imaging agents for amyloid and other targets.
NIR imaging has been used for many years to look at functional parameters in the human brain (for example, saturations of hemoglobin)12,13. Although it is unlikely that NIR imaging will permit imaging of the entire brain, it is possible to image several centimeters below the human skull, potentially enabling, for example, detection of amyloid in the cortex. If amyloid imaging were used to diagnose Alzheimer disease, it is likely that simply detecting plaques in areas of the brain known to be affected by Alzheimer disease would be good enough.
The development of amyloid probes that can be imaged in vivo is almost certain to expedite the preclinical and clinical evaluation of novel Alzheimer-disease therapeutics that target amyloid β. Such ligands may also be useful in the diagnosis of atypical Alzheimer-disease cases. But current clinical diagnosis of Alzheimer disease is reasonably accurate, so it is unlikely that amyloid imaging will become a routine diagnostic modality unless it were relatively quick, safe and inexpensive. With further advances in the technology and ligands, NIR imaging of amyloid may fulfill these criteria14.
There is evidence to suggest that amyloid deposition predates the clinical signs of Alzheimer disease by years or even decades; however, the exact temporal relationship between amyloid deposition and cognitive dysfunction remains to be established. The utility of existing amyloid probes for detecting very early stages of amyloid deposition in the brain of humans has not yet been determined, although most believe that significant improvements in sensitivity will be needed. As it is almost certain that Alzheimer disease will be easier to prevent than treat, a refined version of current amyloid imaging methods may ultimately be the diagnostic tool used to determine both who needs prophylactic treatment and when that treatment should be initiated.