We report increased PiB retention at the sites of CAA-related CMB. PiB DVR values were elevated at CMB sites and declined with increasing distance. The findings support the possibility that vessel wall breakdown and consequent hemorrhage occur preferentially in local “islands” of increased vascular amyloid. Although the absolute difference in DVR between observed and simulated CMB (1.20 vs 1.11) appears small, it represents nearly a doubling in amount of PiB retention above the expected background DVR value of 1. This difference is also of comparable magnitude to reported differences in global PiB DVR between CAA and normal control (0.14) or between AD and normal control (0.37).5
In testing whether these findings were greater than expected by chance, we chose a null hypothesis in which simulated lesions were placed preferentially in those brain regions (occipital, posterior parietal, posterior temporal; Supplemental Figure 1
) favored by actual CAA-related CMB. Differences from this null hypothesis therefore represent specific correlations between CMB and PiB within individual scans rather than a general tendency of CAA to occur in certain brain regions.
Along with CMB, advanced CAA is also associated with white matter lesions,10
increased tissue diffusivity,11
and small cerebral infarction.12
Although these small lesions might appear clinically “silent,” epidemiologic, pathologic, and radiologic data indicate that they can act additively or synergistically with AD pathology to cause cognitive impairment.13,14
The independent contribution of CMB to neurologic dysfunction remains to be determined; some studies point to a possible associations with functional impairment.15, 21
There are several limitations of our study to consider. PiB-PET detects fibrillar amyloid in senile plaques as well as CAA. Since there is no known relationship between plaque amyloid and microbleeding, however, we would expect this effect to act as background noise biasing our results towards the null hypothesis rather than a positive result. A second technical factor is the larger voxel size (i.e. lower spatial resolution) of PET relative to MRI. We would also expect this effect to bias towards a null result, as the large size of the PiB-PET voxels would make it more difficult to detect changing PiB levels in the examined volumes. As our study was cross-sectional, we were unable to determine the cause-effect relationship between increased PiB retention and CMB. The neuropathologic observation that PiB does not label CMB argues, however, that the results do not simply represent nonspecific binding of this agent to blood products. Finally, we note that our results are dominated by the relatively small number of cases with large CMB counts. It is thus unclear whether the results apply to brains with only a few CAA-related CMB.
The implication of our data is that local accumulation of large amounts of vascular amyloid is a necessary step for CAA-related hemorrhage. From this standpoint, the data suggest that strategies to reduce vascular amyloid burdens might be useful for preventing hemorrhage. The results to date from one such strategy—anti-amyloid immunotherapy—have suggested a more complex picture, as both animal and human studies have found instances of increased hemorrhage.16,17
Hemorrhage in association with amyloid immunotherapy may reflect other dynamics besides reduced vascular amyloid burden, including possible transfer of plaque amyloid into vessels.18
or the acute response of amyloid-laden vessels to antibody-mediated clearance.19
A possibility that remains to be tested is whether the long-term result of anti-amyloid immunotherapy would be to reduce vascular amyloid burden and thus diminish the risk of further CAA-related hemorrhage.