In this series of 57 BLSA-NI participants with longitudinal cognitive assessments, we found that higher [11C]PiB retention was associated with steeper trajectories of cognitive decline in the years preceding and concurrent with the PiB evaluation. Specifically, higher [11C]PiB was associated with longitudinal decline in mental status and verbal episodic memory although mean cortical [11C]PiB retention was not related to baseline level of performance on any of the cognitive measures. Six individuals in our sample had mild memory loss by CDR, but only one met consensus criteria for MCI at the time of the study. Associations between higher Aβ burden and decline in mental status and immediate free recall remained significant after exclusion of these individuals, and additionally, decline in executive function was associated with higher [11C]PiB mDVR after exclusion of individuals with mild memory loss. These results provide evidence that in vivo measures of Aβ deposition with PET and [11C]PiB are associated with longitudinal cognitive change even in clinically normal individuals.
Our findings of associations between [11
C]PiB retention and longitudinal decline in verbal episodic memory in clinically normal individuals extend observations of cross-sectional associations reported in some studies,7,16
although we did not find significant relationships between baseline cognitive function and Aβ burden. Higher [11
C]PiB retention in association with greater longitudinal decline in verbal recall is also consistent with the sole longitudinal report,8
which indicated that older adults without dementia who show declines in word list recall are more likely to have higher [11
C]PiB retention compared to individuals with stable clinical status and verbal recall. Furthermore, our findings hold even after excluding individuals who had developed very mild cognitive impairment, suggesting that associations between PiB retention and cognitive change emerge early in the disease process before clinical symptoms are apparent.
Although higher [11
C]PiB retention was associated with decline in verbal recall performance, we observed no significant associations between [11
C]PiB retention and longitudinal change in BVRT performance, a measure of short-term visual memory which declines early in the course of AD.3,12
The lack of association between BVRT performance and [11
C]PiB retention has also been reported in cross-sectional studies.7
It is possible that the lack of an association for BVRT reflects greater sensitivity of this measure to mesial temporal dysfunction and the low level of Aβ deposition visualized by [11
C]PiB in mesial temporal structures. In contrast, the CVLT is sensitive to frontal as well as temporal function, due to its greater dependence on the capacity to organize serial and semantic information. Voxel-based analyses offer support for this explanation, as CVLT associations with [11
C]PiB retention were observed for prefrontal and lateral temporal but not mesial temporal regions. This explanation also receives some support from our observed association between decline in frontal executive function, as measured by Trails B, and [11
C]PiB retention in individuals with CDR = 0 (which reached only trend level in the entire sample, p
< 0.10). Perhaps, future tools for in vivo measurement of the neurofibrillary tangles that are present early in mesial temporal structures would show different patterns of regional correlations across cognitive domains.
Another issue raised by our observations of associations between elevated [11
C]PiB retention and longitudinal declines in verbal episodic memory and executive function is the apparent inconsistency with postmortem findings indicating similar trajectories of longitudinal cognitive performance in clinically normal individuals with and without AD pathology.3
In postmortem studies, we compared antemortem cognitive function in a selected group of individuals who remained clinically stable until death despite AD pathology (asymptomatic AD) with groups of clinically and neuropathologically normal controls and clinically and neuropathologically abnormal AD and MCI. In contrast, our in vivo [11
C]PiB studies include a mixed group of individuals who will remain cognitively normal as well as those who will go on to develop cognitive impairment and AD. An important difference between our autopsy sample and the in vivo study is the age at which Aβ burden was measured. Mean ages (years) in the autopsy sample were 78.6 for controls, 85.8 for cognitively normal subjects with pathology, and 89.4 for subjects with MCI/AD, whereas mean age in the imaging sample was 78.8 years. Thus, participants with elevated Aβ burden in the imaging sample were less likely to have passed fully through the risk period for cognitive impairment, and individuals with elevated [11
C]PiB values and cognitive decline may represent those most likely to develop AD.8
An additional advantage of in vivo imaging of Aβ deposition is that it enables prospective evaluation of individuals with elevated Aβ burden who remain cognitively stable vs those who eventually show more marked cognitive decline and impairment. Such studies may aid in identifying possible protective and compensatory factors that allow some individuals with neuropathology to retain cognitive health.
Limitations of our study include the small sample size and the fact that [11
C]PiB imaging was not possible at baseline cognitive assessment. In addition, the BLSA is a highly educated community dwelling sample, which limits the generality of our findings. However, the incidence of AD in the BLSA12
and the rates of brain changes17
are comparable to other samples. Our study also has several strengths. BLSA-NI participants were followed prospectively for at least 8 years prior to initial [11
C]PiB evaluation, and cognitive status was determined within the context of these longitudinal assessments, with 51 participants rated as CDR = 0 at the time of [11
C]PiB evaluation. While we do not yet know which individuals will eventually develop cognitive impairment, our analyses represent a snapshot in time and the fact that most individuals have had stable cognition over ≥8 years increases the likelihood that the majority of participants will remain clinically normal through the next few years of follow-up. Another strength of our study is that our outcome measures of episodic memory, including the CVLT and BVRT, are not used in determination of diagnostic status. Thus, associations between [11
C]PiB retention and change on these measures are determined independently of clinical determinations of cognitive status.
Continued longitudinal assessments of Aβ deposition in conjunction with cognitive performance and diagnostic status will provide insights into behavioral manifestations and the pathobiology of disease progression at the earliest stages of the neurodegenerative disease process. In vivo imaging of neuropathology has the potential to identify not only risk factors for clinical manifestation of disease, but also to elucidate possible protective and compensatory factors that maintain cognitive health in successful aging.